article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
elliptical galaxies do not show the presence of hii regions and it is not possible to resolve single stars in them in order to measure photospheric abundances .
therefore , most of the information on these objects is obtained from their integrated properties : abundances are derived either through colors or integrated spectra and in both cases the derived information is a complicated measure of metallicity and age ( the well known age - metallicity degeneracy ) .
the most common metallicity indicators are @xmath2 and @xmath1 , as originally defined in faber et al .
( 1977 ; 1985 ) .
population synthesis techniques are adopted to analyze the integrated properties of ellipticals and to derive an estimate of their real abundances .
unfortunately , they contain several uncertainties residing either in incomplete knowledge of stellar evolution or in deficiencies in stellar libraries , as discussed in charlot et al .
( 1996 ) . in recent years
more and more population synthesis models ( bruzual and charlot , 1993 ; buzzoni et al . , 1992 ; bressan et al . , 1994 ;
gibson , 1997 ; bressan et al .
, 1996 ; gibson and matteucci , 1997 ; tantalo et al . , 1998 ) have appeared but the basic uncertainties still remain . in this paper
we want to focus our attention about the comparison between theoretical model results and metallicity indicators . in this framework
we will analyze the relationship between @xmath1 and @xmath0 and its implications for the mechanism of galaxy formation . several authors ( faber et al . 1992 ; worthey et al . 1992 ; carollo et al . 1993 ; davies et al .
1993 ; carollo and danziger 1994 ) , from comparison of the observed indices with synthetic indices , concluded that the average @xmath3_{*}$ ] in giant ellipticals must be larger than the solar value .
this result was also confirmed by the analysis of weiss et al .
( 1995 ) who made use , for the first time , of stellar evolutionary tracks calculated under the assumption of non - solar ratios.the same authors found that the @xmath1 versus @xmath2 relation among nuclei of giant ellipticals is rather flat and flatter than within galaxies . from the flat behavior of @xmath1 vs. @xmath0
the same authors inferred that the abundance of mg should increase faster than the abundance of fe among nuclei of giant ellipticals .
this conclusion is at variance with the predictions of supernova - driven wind models of ellipticals ( arimoto and yoshii , 1987 ; matteucci and tornamb 1987 ) .
in fact , matteucci and tornamb ( 1987 ) showed that , in the framework of the classic wind model for ellipticals , the [ mg / fe ] ratio is a decreasing function of the galactic mass and luminosity .
the reason for this behavior is clear : if fe is mostly produced by the supernovae of type ia , as it seems to be the case in our galaxy ( greggio and renzini 1983a ; matteucci and greggio 1986 ) , whereas mg is mostly originating from supernovae of type ii , then the iron production is delayed relative to that of mg and its abundance should be larger in more massive galaxies which develop a wind later than the less massive ones .
all of this is valid under the assumption that after the onset of a galactic wind star formation should stop or should be negligible , which is a reasonable assumption for elliptical galaxies .
faber et al .
( 1992 ) proposed alternative scenarios to the classic supernova driven wind model , as originally proposed by larson ( 1974 ) .
they suggested three different scenarios all based on the assumption that mg is produced by type ii supernovae and fe is mostly produced by type ia supernovae : i ) a selective loss of metals , ii ) a variable initial mass function ( imf ) and iii ) different timescales for star formation .
these hypotheses were discussed by matteucci ( 1994 ) , who tested them in the context of chemical evolution models . in the hypothesis of the different timescales for star formation matteucci ( 1994 ) suggested that the more massive ellipticals might experience a much stronger and faster star formation than less massive ellipticals leading to a situation where the most massive objects are able to develop galactic winds before the less massive ones .
she called this case `` inverse wind model '' . on the other hand , in the classic wind model of larson ( 1974 )
the efficiency of star formation was the same for all galaxies thus leading to the fact that the galactic wind in more massive systems occurs later than in less massive ones , due to their deeper potential well .
in the models of arimoto and yoshii ( 1987 ) and matteucci and tornamb ( 1987 ) the efficiency of star formation was a decreasing function of galactic mass , based on the assumption that the timescale for star formation is proportional to the cloud - cloud collision timescale which , in turn , is proportional to the gas density .
therefore , since in this monolithic collapse picture the gas density decreases with the galactic mass , the galactic wind was even more delayed for the most massive systems .
matteucci ( 1994 ) proposed , as an alternative , a star formation efficiency increasing with the galactic mass and she justified this assumption by imagining that giant elliptical galaxies , instead of forming through a monolithic collapse of a gas cloud , form by merging of gaseous protoclouds .
the merging process can , in fact , produce higher densities for increasing galactic mass and/or higher cloud - cloud collision velocities resulting in a faster star formation process .
in such a model the galactic wind occurs earlier in massive than in smaller ellipticals thus producing the expected trend of an increasing [ mg / fe ] as a function of galactic mass .
matteucci ( 1994 ) also showed that a variable imf with the slope decreasing with increasing galactic mass and luminosity can produce the same effect without an inverse wind situation .
the reason for that resides in the fact that a flatter imf slope favors massive stars relative to low and intermediate masses , thus favoring mg production over fe production .
however , matteucci ( 1994 ) could not translate the predicted abundances into @xmath0 and @xmath1 since there were no available calibrations for [ fe / h ] versus @xmath1 but only calibrations for @xmath4 $ ] vs. @xmath0 .
therefore she did not compare the predicted abundances with observations .
recently , calibrations for the iron index have become available ( borges et al .
, 1995 ; tantalo et al . ,
1998 ) and therefore in this paper we revisit the whole problem of inferring trends on the real abundances by metallicity indices and we discuss the influence of the calibration relationships , which allow us to pass from indices to abundances , and we show that the inferred trend of mg / fe with galactic mass is not so clear when interpreted in terms of real abundances , thus warning us from drawing any firm conclusion on galaxy formation processes just on the basis of the observed behavior of @xmath1 versus @xmath2 . indices .
the reason for that resides partly in the large spread present in the observational data and partly in the fact that metallicity indices depend not only on the abundances of single elements but also on the ages and on the metallicity distribution ( tantalo et al .
1998 ) of the different stellar populations present in elliptical galaxies .
in section 2 we will discuss the chemical evolution model , in section 3 we will define the average abundances of a composite stellar population , in section 4 we will describe the model results and transform the predicted abundances into indices by means of the most recent metallicity calibrations . finally in section 5
some conclusions will be drawn .
the adopted model of galactic evolution is that outlined by matteucci and gibson ( 1995)(hereafter mg95 ) , where extensive descriptions and references can be found . the evolution of the abundances of several chemical species ( h , he , c , n , o , ne , mg , si and fe ) in the gas is followed , taking into account detailed nucleosynthesis from stars of all masses and sne of types ia , ib , and ii .
we assume that ellipticals can be considered initially as homogeneous spheres of gas with luminous mass in the range @xmath5 m@xmath6 .
a single zone interstellar medium ( ism ) with instantaneous mixing of gas is assumed throughout .
the adopted age for all galaxy models is @xmath7 gyr .
the fundamental equations can be written @xmath8{\rm d}m } \nonumber \\ & & + ( 1-a)\int_{m_{\rm bm}}^{m_{\rm bm}}{\psi(t-\tau_{m})q_{mi}(t-\tau_m)\phi(m ) { \rm d}m } \nonumber \\ & & + \int_{m_{\rm bm}}^{m_{\rm u}}{\psi(t-\tau_m)q_{mi}(t- \tau_m)\phi(m){\rm d}m } , \label{eq : evolution}\end{aligned}\ ] ] where @xmath9 is the volume gas density in the form of an element _ i _ normalized to the initial total volume gas density .
the quantity @xmath10 represents the abundance by mass of an element _ i _ and by definition the summation over all the elements present in the gas mixture is equal to unity .
the various integrals in equation ( 1 ) represent the rates at which sne ( i and ii ) as well as single low and intermediate mass stars and single massive stars restore their processed and unprocessed material into the ism ( for a detailed description of the integrals see mg95 ; matteucci and greggio 1986 ) .
we only remind here that the quantity @xmath11 represents the fraction of a star of mass @xmath12 which is restored in to the ism in the form of an element @xmath13 and therefore contains the nucleosynthesis prescriptions that we assume to be the same as in mg95 .
the star formation rate @xmath14 is given by : @xmath15 normalized to the initial total volume density .
@xmath14 is assumed to drop to zero at the onset of the galactic wind .
the quantity @xmath16 is expressed in units of gyr@xmath17 and represents the efficiency of star formation , namely the inverse of the time scale of star formation .
the values adopted here for @xmath18 are the same as in matteucci ( 1992 ) and arimoto and yoshii ( 1987 ) . in one case
we have adopted the prescription for the * inverse wind model * of matteucci ( 1994 ) .
the difference between the two cases is that in matteucci ( 1994 ) the * classic wind model * assumes that the efficiency of star formation decreases with increasing total galactic mass , as in arimoto and yoshii ( 1987 ) , whereas in the inverse wind model the efficiency of star formation increases as the total galactic mass increases ( similar in spirit to the sfr efficiency parametrization of tinsley and larson 1979 ) leading to a situation in which more massive galaxies experience a galactic wind before the less massive ones .
the quantity @xmath19 represents the lifetime of a star of mass @xmath12 , and is taken from padovani and matteucci ( 1993 ) . for gas to be expelled from a galaxy
the following condition should be satisfied : the thermal energy of the gas heated by sn explosions should exceed the binding energy of the gas ( larson 1974 ) . at this point the gas present in the galaxy is swept away and the subsequent evolution is determined only by the amount of matter and energy which is restored to the ism by the dying stellar generations . in particular , only low mass stars contribute to this evolutionary phase and , among the sne , only sne of type ia ( those sne events whose progenitors have the longest lifetimes ) .
therefore , in order to evaluate the time for the onset of a galactic wind we need to know the energy input from sne and the binding energy of the gas as a function of time .
the total thermal energy of the gas at time t , @xmath20 , and the binding energy of the gas in presence of a diffuse halo of dark matter are calculated as described in matteucci ( 1992 ) .
in particular , @xmath21 is calculated by assuming that @xmath22 of the initial blast wave energy is transferred into the ism as thermal energy by a sn remnant , if the time elapsed from the sn explosion is shorter than a sn remnant cooling time ( cox 1972 ) .
the percentage of transferred energy then decreases as a power law in time @xmath23 for times larger than the cooling time .
this is the same formulation used by arimoto and yoshii ( 1987 ) , matteucci and tornamb ( 1987 ) and mg95 with the exception that we consider a more realistic cooling time ( expressed in years ) which takes into account that the gas density is changing with time and that part of the interstellar gas is in the form of he : @xmath24^{-0.53}\,\ , years .
\label{eq : tc}\end{aligned}\ ] ] other formulations of the energy input from sne can be found in gibson ( 1997 ) and gibson and matteucci ( 1997 ) .
the energetic input from stellar winds in massive stars is ignored , since for normal ellipticals it is negligible compared to the sn thermal energy contribution , as shown by gibson ( 1994 ) .
the galactic wind may last only for few @xmath25 years or continue until the present time depending crucially on the assumptions about the thermal energy of the gas and the potential energy of the gas .
unfortunately , none of these quantities is well known .
the time for the occurrence of the galactic wind , @xmath26 , either increases with the galactic mass as a consequence of the potential well incresing with galactic mass together with the efficiency of star formation decreasing with galactic mass ( classic wind model ) , or decreases with the galactic mass if the efficiency of star formation is strongly increasing with galactic mass ( inverse wind model ) , as can be seen in table 3 , as will be discussed in section 4 .
lllll @xmath27(@xmath28 ) & @xmath18(@xmath29 ) & @xmath30(kpc ) & @xmath31(gyr ) & @xmath32(@xmath28 ) + @xmath33 & 19.0 & 0.5 & 0.131 & @xmath34 + @xmath35 & 14.6 & 1.0 & 0.286 & @xmath36 + @xmath37 & 11.2 & 3.0 & 0.514 & @xmath38 + @xmath39 & 8.6 & 10.0 & 0.955 & @xmath40 + lllll @xmath27(@xmath28 ) & @xmath18(@xmath29 ) & @xmath30(kpc ) & @xmath31(gyr ) & @xmath32(@xmath28 ) + @xmath33 & 19.0 & 0.5 & 0.069 & @xmath41 + @xmath35 & 14.6 & 1.0 & 0.117 & @xmath42 + @xmath37 & 11.2 & 3.0 & 0.403 & @xmath43 + @xmath39 & 8.6 & 10.0 & 0.660 & @xmath44 + lllll @xmath27(@xmath28 ) & @xmath18(@xmath29 ) & @xmath30(kpc ) & @xmath31(gyr ) & @xmath32(@xmath28 ) + @xmath33 & 2 . & 0.5 & 1.200 & @xmath45 + @xmath35 & 5 . &
1.0 & 0.897 & @xmath46 + @xmath37 & 11 . & 3.0 & 0.408 & @xmath47 + @xmath39 & 20 . & 10.0 & 0.205 & @xmath48 + llllll @xmath27(@xmath28 ) & @xmath18(@xmath29 ) & @xmath30(kpc ) & @xmath31(gyr ) & @xmath32(@xmath28 ) & imf slope + @xmath33 & 19.0 & 0.5 & 0.163 & @xmath49 & 2.0 + @xmath35 & 14.6 & 1.0 & 0.237 & @xmath50 & 1.7 + @xmath37 & 11.2 & 3.0 & 0.514 & @xmath38 & 1.35 + @xmath39 & 8.6 & 10.0 & 0.660 & @xmath51 & 0.95 + lllll @xmath27(@xmath28 ) & @xmath18(@xmath29 ) & @xmath30(kpc ) & @xmath31(gyr ) & @xmath32(@xmath28 ) + @xmath33 & 19.0 & 0.5 & 0.073 & @xmath52 + @xmath35 & 14.6 & 1.0 & 0.223 & @xmath53 + @xmath37 & 11.2 & 3.0 & 0.601 & @xmath54 + @xmath39 & 8.6 & 10.0 & 2.205 & @xmath55 + lllll @xmath27(@xmath28 ) & @xmath18(@xmath29 ) & @xmath30(kpc ) & @xmath31(gyr ) & @xmath32(@xmath28 ) + @xmath33 & 4.0 & 0.5 & 0.22 & @xmath56 + @xmath35 & 18.0 & 1.0 & 0.110 & @xmath57 + @xmath37 & 32.0 & 3.0 & 0.090 & @xmath58 + @xmath39 & 75.0 & 10.0 & 0.030 & @xmath59 +
in order to compare model results with observations we should first calculate the average abundances of mg and fe of the composite stellar population of the given galaxies .
the average metallicity or abundance in general which should be compared with the indices should be averaged on the visual light , namely : @xmath60 where @xmath61 is the number of stars in the abundance interval @xmath62 and luminosity @xmath63 . on the other hand
, the real average abundance should be the mass - averaged one , namely : @xmath64 where the subscript @xmath65 refers to the specific time @xmath66 ( the present time ) and @xmath67 is the total mass of stars ever born . here
we will use eq .
( 5 ) in order to compare model results with indices and the reason is that we want to compare our results with previous ones ( matteucci 1994 ; mg95 ) and because for giant ellipticals the difference between the mass - averaged metallicity and the luminosity - averaged one is negligible ( yoshii and arimoto , 1987 ; gibson 1997 ) . on the other hand , in smaller systems the difference between the two abundances is not negligible due to the contribution to the light of low metallicity red giants ( greggio 1996 ) .
we will show in section 4 that in the cases studied here @xmath68 .
in this section we will show the results of several models and we will convert , by means of the available calibrations , the average stellar abundances of mg and fe , @xmath69_{*}$ ] and @xmath70_{*}$ ] , as predicted for elliptical galaxies of different mass , into the metallicity indices @xmath2 and @xmath1 , respectively .
let us discuss first the metallicity calibrations .
relations linking the strength of metallicity indices to real abundances can be either empirical or theoretical . in the past few years
several attempts have been made to calibrate the strength of @xmath2 against [ fe / h ] which has always been considered as the measure of the `` metallicity '' in stars .
it should be said that this is not entirely correct since we know that mg does not evolve in lockstep with iron in the solar neighborhood nor in elliptical galaxies , due to the different timescales of production of these two elements .
it would be much better to calibrate @xmath0 versus [ mg / h ] in order to avoid confusion .
calibrations of @xmath0 versus @xmath1 are from mould ( 1978 ) , burstein ( 1979 ) , peletier ( 1989 ) , buzzoni et al .
( 1992 ) , worthey et al .
( 1992 ) . in all of these calibrations
the ratio [ mg / fe ] is assumed to be solar , at variance with the indication arising from population synthesis models showing an overabundance of mg relative to iron in the nuclei of giant ellipticals ( faber et al .
1992 ; worthey et al . 1992
; davies et al . 1993 ; weiss et al .
1995 ) .
more recently , barbuy ( 1994 ) , borges et al .
( 1995 ) and tantalo et al .
( 1998 ) took into account non solar ratios of [ mg / fe ] in their calibrations .
in addition , some of them ( borges et al .
1995 ; tantalo et al .
1998 ) produced synthetic @xmath1 indices thus allowing us to calibrate [ fe / h ] also against @xmath1 .
this allows us to transform [ fe / h ] in to @xmath0 and @xmath1 , although many uncertainties are involved in this exercise , mainly because , in this way , the derivations of @xmath0 and @xmath1 are not independent .
we run several models , in particular : model i , which is the classic wind model , as described in mg95 , with a salpeter ( 1955 ) imf ( namely an imf with power index x=1.35 over a mass range @xmath71 ) ; model ii , which is the classic wind model with the arimoto and yoshii ( 1987 ) imf ( namely an imf with power index x=0.95 over the same mass range of the salpeter one ) ; model iii , which is the equivalent of the inverse wind model , as described in matteucci ( 1994 ) with the arimoto and yoshii ( 1987 ) imf ; model iv which is the equivalent of the model with variable imf , as described in matteucci ( 1994 ) , which assumes that ellipticals of smaller mass have a steeper imf than the more massive ones .
it is worth noting that the slope of the imf is kept constant inside a galaxy .
in particular , we vary the slope of the imf from the salpeter one to the arimoto and yoshii one passing from a galaxy with initial luminous mass of @xmath72 to a galaxy with @xmath73 . this particular assumption can reproduce the observed tilt of the fundamental plane seen edge - on , namely the increase of m / l versus l as observed by bender et al .
( 1992 ) .
model v assumes a time - variable imf as suggested by padoan et al .
( 1997 ) and will be discussed in a forthcoming paper . in this formulation
the imf slope varies as a function of gas density and gas velocity dispersion , favoring the formation of massive stars at early epochs .
model vi assumes a constant imf with a slope x=0.8 and a star formation efficiency which varies more strongly with the luminous mass than in model iii .
the slope x=0.8 is the limiting slope that we can accept to obtain a realistic @xmath74 ratio for ellipticals , as discussed in padovani and matteucci ( 1993 ) .
the model parameters are described in tables 1 - 6 where we list the luminous masses in column 1 , the star formation efficiency ( in units of @xmath29 ) in column 2 , the effective radius ( in units of kpc ) in column 3 , the time for the occurrence of the galactic wind ( in gyr ) in column 4 and the final galactic luminous mass in column 5 . for model iv
is shown also the slope of the imf in column 6 .
lllll @xmath75(@xmath28 ) & @xmath70_\ast$](dex ) & @xmath3_\ast$](dex ) & @xmath0(mag ) & @xmath1( ) + @xmath76 & -0.395 & 0.262 & 0.217 & @xmath77 + @xmath78 & -0.190 & 0.224 & 0.247 & @xmath79 + @xmath80 & -0.095 & 0.187 & 0.257 & @xmath81 + @xmath82 & -0.007 & 0.130 & 0.260 & @xmath83 + lllll @xmath75(@xmath28 ) & @xmath70_\ast$](dex ) & @xmath3_\ast$](dex ) & @xmath0(mag ) & @xmath1( ) + @xmath76 & -0.833 & 0.397 & 0.152 & @xmath84 + @xmath78 & -0.034 & 0.329 & 0.298 & @xmath85 + @xmath80 & 0.388 & 0.288 & 0.368 & @xmath86 + @xmath82 & 0.502 & 0.264 & 0.383 & @xmath87 + lllll @xmath75(@xmath28 ) & @xmath70_\ast$](dex ) & @xmath3_\ast$](dex ) & @xmath0(mag ) & @xmath1( ) + @xmath76 & 0.197 & 0.262 & 0.326 & @xmath88 + @xmath78 & 0.334 & 0.274 & 0.355 & @xmath89 + @xmath80 & 0.384 & 0.290 & 0.368 & @xmath90 + @xmath82 & 0.317 & 0.308 & 0.360 & @xmath91 + lllll @xmath75(@xmath28 ) & @xmath70_\ast$](dex ) & @xmath3_\ast$](dex ) & @xmath0(mag ) & @xmath1( ) + @xmath76 & -1.450 & 0.149 & 0.002 & @xmath92 + @xmath78 & -0.834 & 0.169 & 0.125 & @xmath93 + @xmath80 & -0.095 & 0.187 & 0.257 & @xmath81 + @xmath82 & 0.502 & 0.264 & 0.368 & @xmath87 + lllll @xmath75(@xmath28 ) & @xmath70_\ast$](dex ) & @xmath3_\ast$](dex ) & @xmath0(mag ) & @xmath1( ) + @xmath76 & -0.550 & 0.339 & 0.200 & @xmath94 + @xmath78 & 0.115 & 0.253 & 0.309 & @xmath95 + @xmath80 & 0.267 & 0.136 & 0.308 & @xmath96 + @xmath82 & 0.291 & -0.058 & 0.263 & @xmath97 + lllll @xmath75(@xmath28 ) & @xmath70_\ast$](dex ) & @xmath3_\ast$](dex ) & @xmath0(mag ) & @xmath1( ) + @xmath76 & -0.099 & 0.340 & 0.287 & @xmath98 + @xmath78 & -0.091 & 0.365 & 0.294 & @xmath99 + @xmath80 & -0.023 & 0.368 & 0.308 & @xmath100 + @xmath82 & -0.005 & 0.390 & 0.316 & @xmath101 + we then calculate the average @xmath70_{*}$ ] and @xmath3_{*}$ ] for the stellar component of ellipticals by using eq . ( 6 ) and finally we transform these abundances to observed @xmath0 and @xmath1 line indices . in tables 7 - 12 we show the results for different models and for the calibration of tantalo et al .
in particular , in tables 7 - 12 we show the luminous mass in column 1 , the @xmath70_{*}$ ] in the second column , the @xmath3_{*}$ ] in column 3 and in column 4 and 5 the @xmath0 and the @xmath1 indices , respectively . as already said , only the calibrations of tantalo et al .
( 1998 ) and borges et al .
( 1995 ) allow us to transform [ fe / h ] into @xmath1 and therefore allow us to compare model results with the data showing the behavior of @xmath0 vs. @xmath1 among nuclei of giant ellipticals .
in particular , starting from the synthetic indices of tantalo et al .
( 1998 ) calculated for a fixed [ mg / fe ] and a fixed [ fe / h ] we derived calibration relationships of the type : @xmath102 , [ mg / fe])\ ] ] @xmath103 , [ mg / fe])\ ] ] which allowed us to derive the indices for any [ fe / h ] and [ mg / fe ] .
the calibrations we have adopted are : @xmath104 + \nonumber \\ + ( 0.153 + 0.120[mg / fe])\cdot [ fe / h]\end{aligned}\ ] ] @xmath105 + \nonumber \\ ( 1.654 - 0.307[mg / fe])\cdot [ fe / h]\end{aligned}\ ] ] in fig .
1 we show the metallicity indices obtained by means of the already mentioned calibrations compared with the data ( gonzalez 1993 ; worthey et al .
1992 ; carollo and danziger 1994a , b ) . as one
can easily see the data present a large spread , mostly due to the uncertainties in deriving the @xmath1 indices .
in particular , in fig 1a we show the observed and predicted behavior of @xmath1 vs. @xmath0 when model i is adopted .
the bestfit to the data implies the following relation , @xmath106 , and is indicated in the figure .
however , the spread in the data is large and this prevents us from drawing strong conclusions about a possible trend . the dotted lines in fig .
1a represent the predictions of model i obtained by means of the calibrations described before and adopting the same [ mg / fe ] ratio as predicted by the models , as one can see in table 7 .
the agreement with the observed trend is not so good , showing that the slope of the predicted relation is steeper than that shown by the data and that the predicted @xmath0 values do not cover the entire range in @xmath0 .
this is mostly due to the assumed imf since model ii , which assumes a flatter imf , predicts values for @xmath0 which cover the whole range ( see fig .
2a ) . in fig .
1b we show the predicted and observed mass - metallicity ( @xmath0 ) relationship .
the data are from carollo et al .
the best - fit to these data indicate @xmath107 , where @xmath108 is the total galactic mass ( dark+luminous ) .
the classic wind model recovers the slope of the @xmath0 mass relation , but with a zeropoint offset of @xmath109 with respect to the observed distribution . on the other hand , the classic wind model with the arimoto and yoshii ( 1987 ) imf ( model
ii ) , as shown in fig.2b , predicts a slope much steeper than the observed one , although it agrees better than model i with the @xmath1 vs. @xmath0 relation shown in fig .
it is worth noting that the arimoto and yoshii ( 1987 ) imf well reproduces the abundances in the intergalactic medium ( mg95 ; gibson 1997 ; gibson and matteucci 1997 ) .
it is worth noting that in figures 1a and 2a and in all the others we show also the relation between real abundances predicted by our models .
the relation between @xmath70_{*}$ ] and @xmath69_{*}$ ] , arbitrarily translated in the plot of @xmath1 versus @xmath0 , is indicated by the dashed lines .
this is done only with the purpose of comparing the slope of the relation between real abundances with that of the relation between indices and they are very similar , indicating that the adopted calibration does not modify the predicted relation between mg and fe abundances .
one of the main reasons for that is the adopted calibration which accounts for the right @xmath3_{*}$ ] ratio for each galaxy . in fig.3a
, b we show the predictions of the inverse wind model of matteucci ( 1994 ) which predicts a stellar @xmath3_{*}$ ] increasing with galactic mass .
the slope of the @xmath1 versus @xmath0 relation is in better agreement than in the previous models , and the slope of the @xmath0 vs. mass relation is also acceptable although the absolute values of the indices are too high . in fig.4a , b we show the results of model iv with a variable imf from galaxy to galaxy , which also predicts increasing @xmath3_{*}$ ] ratios with galactic mass .
the agreement with the @xmath1 vs. @xmath0 data is marginally acceptable , but the slope of the mass - metallicity relation is too steep and the predicted absolute values of @xmath0 are too low .
the low absolute values of @xmath0 are due to the fact that we used slopes steeper than the salpeter ( 1955 ) one for the less massive galaxies and such slopes are not suitable for elliptical galaxies ( see mg95 ) since they predict too low metallicities .
however , other numerical experiments , where we used a variable imf but with flatter slopes for each galactic mass ( from x=1.4 in low mass galaxies to x=0.8 in high mass galaxies ) , have shown that there is a negligible difference in the predicted @xmath1 vs. @xmath2 relation while the mass - metallicity relation gets worse . in fig.5a , b we show the predictions of model v calculated with the time - variable imf as suggested by padoan et al .
( 1997 ) and adapted to elliptical galaxies .
the slope of this imf is decreasing with time thus favoring massive stars at early epochs . a similar although more complex formulation of the padoan et al .
( 1997 ) imf has been recently adopted by chiosi et al .
the model behaves like a classic wind model , in the sense that the galactic wind occurs first in small galaxies and later in the more massive ones .
concerning the predicted indices , fig.5a shows that the @xmath0 decreases for massive objects , due to the fact that the imf in these galaxies is less biased towards massive stars than in smaller systems .
this is , in turn , due to the fact that the slope of the imf is inversely proportional to the gas density which is lower in more massive objects .
this model predicts a sort of bimodal behavior for the @xmath0 vs. mass relation and it does not fit the data better than the other models . finally , in fig.6a , b the predictions of model vi are shown .
at variance with all the previous models , model vi seems to reproduce well the observed slope of the @xmath1 vs. @xmath2 relation as well as the @xmath0- mass relation .
the main problem with this model is the fact that the predicted ranges of @xmath0 and @xmath1 are too narrow compared to the observations , especially the range of @xmath1 .
another potential problem of this model is the predicted @xmath74 ratio which is @xmath110 for each galaxy mass .
this is a high value for ellipticals unless one believes in a hubble constant @xmath111 , as discussed in padovani and matteucci ( 1993 ) . in figs .
7 and 8 we show the plot of the mass - metallicity ( @xmath0 ) relation as predicted by model i and model ii , obtained under different assumptions about the calibrating formula .
as one can see , some of the calibrations give similar results such as those of worthey et al .
( 1992 ) , casuso et al .
( 1996 ) and buzzoni et al .
these calibrations have in common the use of a solar ratio for [ mg / fe ] ( [ mg / fe]=0 ) . on the other hand , the values of the indices obtained by using the calibrations of barbuy ( 1994 ) and tantalo et al .
( 1998 ) which adopt non - solar ratios , differ from the others and between themselves .
it is worth noting that the use of different calibrations may lead even to different slopes for the @xmath0-@xmath112 relationship .
one criticism that could in principle be moved to the results discussed before is that we adopted mass - averaged metallicities and not luminosity - averaged metallicities , as it should be the case . in fig .
9 we show the indices obtained by the luminosity- and mass - averaged metallicities calculated for the results of model ii , when the calibration of tantalo et al .
( 1998 ) is applied .
the luminosity - averaged metallicities , computed with the photometric model of gibson ( 1997 ) , are systematically slightly lower than the others and the difference is larger for smaller galaxies , as expected
. however , the slope is the same in the two cases , showing that the use of mass - averaged metallicity for this kind of analysis is quite justified .
in this paper we have discussed the relation between metallicity indices , such as @xmath0 and @xmath1 , and total mass in nuclei of ellipticals and its implications in terms of models of formation and evolution of elliptical galaxies . in order to do that we have transformed the average abundance of fe in the composite stellar population of the galaxy , as predicted by different models of chemical evolution , into @xmath0 and @xmath1 indices by means of the available calibrations .
we have shown the results of classic wind models for ellipticals , such as those discussed by arimoto and yoshii ( 1987 ) and matteucci and tornamb ( 1987 ) , as well as the results of models with variable imf from galaxy to galaxy and with galactic winds occurring first in the more massive systems , implying that these systems are older than the less massive ones .
we have found that it is not possible to establish clearly which kind of model should be preferred , first of all because of the large spread present in the data .
moreover , little difference is found in the predicted indices of models which predict a @xmath3_{*}$ ] either increasing or decreasing with galactic mass , although the data seem to suggest an increase of this ratio with galactic mass larger than predicted by any of the models . on this basis
, the classic wind model can not be considered worse than the other models .
actually , the classic wind model with a flat constant imf seems to be the only one which can reproduce the whole range of the observed indices .
however , if we isolate the data from gonzalez ( 1993 ) and do not consider the others , then in order to reproduce the flat slope of the @xmath1 versus @xmath0 relation , as given by the best - fit of the data , one should assume that fe among the nuclei of ellipticals is almost constant whereas mg increases from less massive to more massive nuclei .
this is not achieved by any of the models presented here since it would require quite `` ad hoc '' assumptions especially concerning the type ia sne . from the numerical experiments performed in this paper
we can say that a model which explains at the same time the mass - metallicity and the iron - magnesium relation requires an inverse wind situation , with a strong increase of the star formation efficiency with galactic mass ( i.e. model vi ) , rather than a variable imf from galaxy to galaxy , and an imf with a slope @xmath113 . however , a model of this type is not able to reproduce the observed ranges of @xmath1 and @xmath0 .
we have also calculated models where amount and concentration of dark matter increases , compatibly with the formulation of the potential energy of the gas , with decreasing galactic luminous mass ( see persic et al .
1996 ) , with the net result of obtaining an `` inverse wind '' situation .
the results are very similar to those of model iii .
therefore , to obtain a better agreement with observations one should invoke also in this case an increase of the star formation efficiency with galactic mass .
this would certainly flatten the @xmath1 vs. @xmath0 relation but it would further shrink the ranges of the predicted indices .
in fact , both an increasing star formation efficiency and a decreasing importance of dark matter with increasing luminous galactic mass can be viable solutions to achieve the situation of more massive ellipticals being older than less massive ones . in conclusion
, it is quite important to establish the value of [ mg / fe ] from the observational point of view since abundance ratios , such as [ mg / fe ] , represent an important diagnostic to infer ages in galaxies , due to the different timescales for the mg and fe production .
generally , a high [ mg / fe ] is interpreted as a young age and the upper limit for the age is given by the time at which the chemical enrichment from type ia sne starts to become important .
this timescale depends not only on the assumed progenitors of type ia sne but also on the star formation history of the galaxy considered ( see matteucci 1997 ) and for giant elliptical galaxies this timescale is of the order of @xmath114 years and in any case it can not be larger than 1 gyr also for smaller systems .
this is at variance with what stated by kodama and arimoto ( 1997 ) who claim , on the basis of results concerning our galaxy ( yoshii et al .
1996 ) , that this timescale is of the order of 1.5 - 2.5 gyr .
this is indeed true for our galaxy where the star formation history has been quite different than in ellipticals and it had been already pointed out in greggio and renzini ( 1983b ) and in matteucci and greggio ( 1986 ) .
this is a quite important point , both for the galactic chemical enrichment and for the predictions about sn rates at high redshift .
therefore , an enhanced [ mg / fe ] indicates that the process of galaxy formation must have been very fast thus favoring a monolithic collapse scenario rather than a merging scenario . in this framework
, a [ mg / fe ] ratio higher in more massive ellipticals than in less massive ones could be interpreted as due to their faster formation and evolution ( see matteucci 1994 ; bressan et al .
1996 ) .
an independent way of estimating the ages of ellipticals , where for ages we intend the time elapsed from the last star formation event , is to study the @xmath115 index .
this index is , in fact , related to the age of the dominant stellar population , since it gives a measure of the turn - off color and metallicity .
it can therefore be used to solve the age - metallicity degeneracy .
bressan et al .
( 1996 ) , by analyzing the @xmath115 and other physical parameters in the sample of ellipticals observed by gonzalez ( 1993 ) , concluded that massive galaxies should have stopped forming stars before less massive ones , in agreement with the results of the inverse wind model discussed here .
finally , we would like to point out that both models with a salpeter imf and a variable imf have a potential problem in reproducing high [ @xmath116/fe ] ratios in the intracluster medium ( icm ) , as shown by their low average @xmath117_*$ ] ratios ( see tables 7 - 12 ) .
therefore , in agreement with mg95 and gibson and matteucci ( 1997 ) we conclude that a flat imf is required to explain the high [ @xmath116/fe ] ratios , as found by asca observations ( mushotzsky 1994 ) .
brocato , e. , matteucci , f. , mazzitelli , i. & tornamb , a. : 1990 , ap.j . * 349 * , 458 bruzual , g. , charlot , s. 1993 ap .
j. * 405 * , 538 burstein , d , 1979 ap .
j. * 232 * , 74 buzzoni , a. gariboldi , g. , mantegazza , l. 1992 a.j . * 103 * , 1814 carollo , c.m . , danziger , i.j . ,
buson , l. : 1993 , mnras * 265 * , 553 carollo , c.m . , danziger , i.j .
1994a , mnras * 270 * , 523 carollo , c.m . , danziger i.j .
1994b , mnras * 270 * , 743 casuso , e. , vazdekis , a. , peletier , r. , beckman , j.e .
1996 ap.j . * 458 * , 533 charlot , s. , worthey , g. , bressan , a. 1996 ap .
j. * 457 * , 625 chiosi , c. , bressan , a.,portinari , l. , tantalo , r. 1998 , a.a . in press cox ,
1972 ap.j .
* 178 * , 159 davies , r.l . , sadler , e.m .
& peletier , r. : 1993 , mnras * 262 * , 650 faber , s.m . , worthey , g. & gonzalez j.j .
1992 , in _ iau symp .
n.149 _ , eds . b. barbuy and a. renzini , p. 255
faber , s.m . , burstein , d. , dressler , a. 1977 , a.j . *
82 * , 941 faber , s.m . ,
freiel , e.d . ,
burstein , d. , gaskell , c.m .
1985 , ap.j .
57 * , 711 gibson , b .. , k. 1994 mnras * 271 * , l35 gibson , b.k .
1997 , mnras * 290 * , 471 gibson , b.k .
, matteucci , f. , 1997 .
mnras , * 291 * , l8 gonzalez , j.j .
1993 phd thesis , university of california , santa cruz greggio , l. , 1996 , mnras * 285 * , 151 greggio , l. , renzini , a. 1983a , in _
`` the first stellar generations '' _ , mem . soc .
astron . it . ,
54 , p.311 greggio , l. , renzini , a. 1983b a. a. * 118 * , 217 kodama , t. , arimoto , n. 1997 , a.a . * 320 * , 41 larson , r.b . 1974 mnras * 169 * , 229 matteucci , f. 1992 ap.j . *
397 * , 32 matteucci , f. 1994 a.a .
* 288 * , 57 matteucci , f. gibson , b.k .
1995 a.a . * 304 * , 11 matteucci , f. , greggio , l. : 1986 , a.a .
* 154 * , 279 mould , j.r .
1978 ap.j . * 220 * , 434 mushotzsky , r. 1994 in `` clusters of galaxies '' ed .
f. durret et al .
editions frontieres p.167 nomoto , k. , thielemann , f. k. & yokoi , k. : 1984 , ap.j . * 286 * , 644 padoan , p , nordlund , a.p . , jones , b.j.t .
1997 , mnras * 288 * , 145 padovani , p. , matteucci , f. 1993 ap.j . *
416 * , 26 peletier , r. , 1989 , phd thesis , university of groeningen persic , m. , salucci , p. , stel , f. 1996 mnras * 281 * , 27 renzini , a. , voli , m. : 1981 , a.a . * 94 * , 175 salpeter , e. e. : 1955 , ap.j . * 121 * , 161 tantalo , r. , chiosi , c. , bressan , a. 1998 a.a . in press thielemann , f.k . , nomoto , k. , hashimoto , m. 1993 _ origin and evolution of the elements _ , ed .
n. prantzos et al . , cambridge univ .
press p. 297 tinsley , b.m . ,
larson , r.b .
1979 mnras * 186 * , 503 weiss , a. , peletier , r. , & matteucci , f . : 1995 a.a . * 296 * , 73 worthey , g. 1994 ap.j . suppl . * 95 * , 107 worthey , g. , faber , s.m . & gonzalez , j.j . : 1992 , ap.j . *
398 * , 73 woosley , s.e . , weaver , t.a .
. astrophys .
* 24 * , 205 woosley , s. e. : 1987 , in _
`` nucleosynthesis and chemical evolution '' _ , 16th saas fee course , eds .
b. hauck et al .
( geneva : geneva observatory ) , p.1 yoshii , y. , arimoto , n. 1987 a.a . * 188 * , 13 yoshii , y. , tsujimoto , t. , nomoto , k. 1996 , ap.j . * 462 * , 266 | we revisit the problem of the flat slope of the @xmath0 versus @xmath1 relationship found for nuclei of elliptical galaxies ( faber et al .
1992 ; worthey et al . 1992 ; carollo et al . 1993 ; davies et al . 1993 ) , indicating that the mg / fe ratio should increase with galactic luminosity and mass .
we transform the abundance of fe , as predicted by classic wind models and alternative models for the chemical evolution of elliptical galaxies , into the metallicity indices @xmath0 and @xmath1 , by means of the more recent index calibrations and show that none of the current models for the chemical evolution of elliptical galaxies is able to reproduce exactly the observed slope of the @xmath1 versus @xmath0 relation , although the existing spread in the data makes this comparison quite difficult .
in other words , we can not clearly discriminate between models predicting a decrease ( classic wind model ) or an increase of such a ratio with galactic mass .
the reason for this resides in the fact that the available observations show a large spread due mostly to the errors in the derivation of the @xmath1 index . in our opinion
this fact prevents us from drawing any firm conclusion on the behaviour of mg and fe in these galaxies .
moreover , as already shown by other authors , one should be careful in deriving trends in the real abundances just from the metallicity indices , since these latter depend also on other physical parameters than the metallicity .
this is an important point since abundance ratios have been proven to represent strong constraints for galaxy formation mechanisms . #
1to 0pt#1 |
since the pioneering work of kaplan and nelson @xcite , that predicted large attractive antikaon self - energies in nuclear matter , speculations about strange condensates in neutron stars and relativistic heavy - ion collisions have raised the interest in kaon - nuclear physics @xcite .
these ideas have been supported by the data on kaonic atoms that indicate a strong attractive potential at zero antikaon momentum @xcite . on the other hand , also the data on @xmath8 production from heavy - ion collisions
@xcite have been interpreted in terms of a dropping antikaon mass at finite density @xcite . whereas the analysis @xcite of the kaonic atom data suggests an attractive antikaon potential of @xmath14180 mev at normal nuclear density , the data on @xmath8 production from heavy - ion collisions lead to a potential @xmath1480@xmath15120 mev @xcite .
this discrepancy might be attributed @xcite to the momentum dependence of the antikaon potential , since the kaonic atoms explore stopped antikaons with @xmath160 , while the heavy - ion experiments have probed the range 300@xmath17600 mev / c . among the available experimental methods
the elastic and inelastic hadron - nucleus scattering are a traditional way @xcite to study the optical potential .
moreover , by measuring the @xmath18 scattering at different projectile momenta one can also evaluate the momentum - dependence of the potential . here
we study the possibility to measure the @xmath8 optical potential by elastic @xmath0 scattering , analyze the available data and discuss the perspectives of @xmath8-meson scattering experiments at low and high momenta .
the elastic scattering of an energetic hadron from a nucleus can be described by the glauber theory of diffractive scattering as formulated in the boulder lectures in theoretical physics @xcite . here
we refer to the final results from ref .
@xcite and present the formulas relevant for the further calculations and discussions .
when neglecting the coulomb distortion of the projectile the amplitude for the elastic scattering of the particle from the nucleus is given as @xmath19\,\right)\ , b \
, db , \label{el1}\ ] ] where @xmath20 is the momentum transfered from the projectile to the nucleus , @xmath21 is the missile momentum , while @xmath22 is the zeroth order bessel function .
the phase shift @xmath23 may be approximated by @xmath24 where @xmath6 is the complex amplitude for the forward scattering of the projectile on a proton or neutron in the nucleus , while @xmath25 is the nuclear density distribution . in principle , @xmath6 should be given by the ( density dependent ) in - medium amplitude , but within the impulse approximation @xcite it can be approximated by the forward scattering amplitude in free space . via the optical theorem , furthermore , the imaginary part of @xmath6 can be expressed in terms of the total particle - nucleon cross section @xmath26 as @xmath27 introducing @xmath28 the phase ( [ shift1 ] ) might be rewritten as @xmath29 the experimental results on the elastic differential cross section given by @xmath30 ( with @xmath31 ) can be used for the evaluation of the cross section @xmath26 and the ratio of the real to imaginary forward scattering amplitude @xmath5 .
recall , that @xmath26 and @xmath5 might differ from the relevant values given in free space ; a question that has to be discussed separately .
neglecting such modifications for a while , the traditional glauber approximation thus allows to evaluate the in - medium parameters describing the interaction of the projectile with a finite nucleus .
furthermore , the coulomb correction may be taken into account by adding the coulomb phase shift to the nuclear phase as proposed in refs .
the coulomb corrected nuclear elastic scattering amplitude then is given by @xmath32 \nonumber \\
\times \left(1{-}\exp[i\chi_n(b){+}i\chi(b)\,]\,\right)\,b \,db .
\label{el2}\end{aligned}\ ] ] the amplitude @xmath33 and the phase shift @xmath34 describe the scattering by a point charge @xmath35 , @xmath36 , \hspace{11 mm } \chi_c(b){=}2\,\xi\ , ln ( kb),\ ] ] where @xmath37 the sign of @xmath38 is given by the attractive ( negative ) or repulsive ( positive ) interaction of the projectile with the coulomb field . the phase shift @xmath39 in eq .
( [ el2 ] ) is given by @xmath40{- } \sqrt{1-\tau^2}\right ) r^2\,dr , \label{shift4}\ ] ] with @xmath41 while @xmath42 is the charge distribution .
we assume the charge density in eq .
( [ shift4 ] ) to be proportional to the nuclear density function in eq .
( [ shift1 ] ) , which is normalized to the total number of nucleons @xmath43 . for 4@xmath4416 nuclei - to a good approximation - it is given by @xcite @xmath45 \exp{\left[-r^2/r^2\right ] } , \label{light}\ ] ] with @xmath46 fm for @xmath12 .
for @xmath4716 nuclei the nuclear density is replaced by the wood - saxon distribution @xmath48 } \label{heavy}\ ] ] with @xmath49 we note that we do not account for a screening phase and discard nucleus recoil corrections @xcite .
the data on the differential cross section for the @xmath8-meson elastic scattering by @xmath12 and @xmath50 at an antikaon momentum of 800 mev / c have been collected by marlow et al .
@xcite and fitted by eq .
( [ el2 ] ) in order to evaluate the cross section @xmath26 and real to imaginary ratio @xmath5 .
we here use the minuit @xcite procedure and the @xmath51 method .
the solid lines in figs .
[ elastic1],[elastic2 ] show the result from the minimization procedure together with the experimental results @xcite .
the parameters @xmath26 and @xmath5 as well as the reduced @xmath51 are listed in table [ tab1 ] . the dashed lines in figs .
[ elastic1],[elastic2 ] show the calculations with @xmath26 and @xmath5 taken from the @xmath3 and @xmath4 interactions in free space and averaged over the number of protons and neutrons in the target as described below .
we note , that although the calculations with the free space variables provide a better description in the vicinity of the diffractive minima , they lead to a substantially worse @xmath52 as shown in table [ tab1 ] .
one of the crucial questions now is the sensitivity of the data to the sign and the magnitude of the ratio @xmath5 .
it is known @xcite that - while @xmath26 is given by the absolute value of the nuclear cross section - the ratio @xmath5 may be fixed by the differential cross section at the diffractive minima .
actually , any effect providing a complex correction of the nuclear scattering amplitude @xmath53 influences the filling of the diffractive minima @xcite .
[ elastic1a ] shows the @xmath54 elastic differential cross section calculated for ratios @xmath5=-0.1 and @xmath5=-0.315 .
the latter value for @xmath5 was taken to be equal in magnitude to the parameter evaluated by the minimization procedure , but by substituting an opposite sign in order to demonstrate the sensitivity of the data to the sign of @xmath5 .
.[tab1 ] the @xmath1 total cross section @xmath26 , ratio @xmath5 of the real to imaginary part of the forward @xmath1 scattering amplitude and @xmath52 evaluated from the data on @xmath0 elastic scattering and that taken from the interaction in free space . [ cols="<,^,^,^,^ " , ] now fig .
[ elastic1a ] illustrates that the absolute value of the ratio @xmath5 might be reasonably fixed by the data . the calculations with @xmath5=-0.315 provide a @xmath52=2.3 , which is worse as compared to the value listed in table [ tab1 ]
indeed , the data are sensitive to both sign and magnitude of the ratio @xmath5 when taking into account the coulomb scattering .
this method has been used before for the evaluation of the real to imaginary ratio @xmath5 in case of the antiproton - nucleon forward scattering amplitude in refs .
@xcite .
since the inclusion of the coulomb correction is important for the determination of the sign of @xmath5 , we also have performed calculations taking into account only the nuclear phase shift .
the results of this limit are shown in fig .
[ elastic1b ] for @xmath8 scattering from @xmath12 and @xmath50 together with the data @xcite .
the parameters @xmath26 and @xmath5 ( extracted from the data ) as well as the associated @xmath52 are also shown in fig .
[ elastic1b ] . note , that when neglecting the coulomb correction the nuclear data become insensitive to the sign of @xmath5 ; furthermore , for @xmath5=0 the differential cross section in the diffractive minima will be zero .
from the analysis presented above it is clear , that without the coulomb scattering it is not possible to reproduce the data at low @xmath20 . although there are not much data available at small scattering angles , the coulomb effect by obvious reason must be included in the data analysis .
we also note that the calculations with the coulomb effect yield the minimal @xmath52 .
we conclude that the data @xcite on elastic @xmath8-meson scattering from @xmath12 and @xmath50 nuclei are sufficiently sensitive to the evaluation of the in - medium total @xmath1 cross section @xmath26 and the ratio @xmath5 of the real to imaginary part of the in - medium forward scattering amplitude @xmath6 .
the resulting parameters are averaged over the proton and neutron numbers of the target . in the following these in - medium parameters
will be compared to those extracted from the @xmath1 interaction in free space .
an important question in @xmath0 scattering is to what extent one can probe the nuclear interior .
obviously , for very peripheral scattering the values for @xmath26 and @xmath5 should be sufficiently close to those values given by the @xmath1 interaction in free space .
indeed , eq . ( [ el1 ] ) indicates that elastic scattering is a peripheral process @xcite .
this is illustrated by fig .
[ elastic7]a ) , where the differential cross section for @xmath8-meson scattering from @xmath12 at @xmath56=0 for an antikaon momentum of 800 mev / c is shown as a function of the impact parameter @xmath57 and the coordinate @xmath58 along the beam axis ( see eq .
( [ el1 ] ) ) .
the calculations show that the scattering is predominantly peripheral . to obtain the density probed by @xmath8 scattering at this momentum we show ( solid line in fig .
[ elastic7]b ) ) the density profile for @xmath12 given by eq .
( [ light ] ) .
the hatched histogram in fig .
[ elastic7]b ) indicates the @xmath54 scattering profile distribution , which was calculated by taking the @xmath54 elastic differential cross section at @xmath59=800 mev / c and @xmath20=0 at given coordinates @xmath57 and @xmath58 and by integrating over the contour @xmath60 .
the scattering profile is normalized to the tail of the density function @xmath61 .
the maximal nuclear density that might be tested by @xmath8 elastic scattering from @xmath12 is about @xmath140.1 @xmath62 , that should be compared to the saturation value @xmath10=0.16 @xmath62 .
the average densities probed by the @xmath8 scattering are in the range of @xmath10/3 . similar conclusions also hold for the @xmath50 nucleus .
this is demonstrated in fig .
[ elastic7a ] which shows the calculations of the reaction zone and the profile functions for @xmath8 scattering from @xmath50 .
now the in - medium modification of the total cross section @xmath26 and the ratio @xmath5 of the real to imaginary part of the forward scattering amplitude and their difference from those values in free space might be discussed for the baryon densities probed in @xmath0 scattering .
the forward @xmath8 scattering amplitude on the proton and the neutron has been calculated in our previous study @xcite . here
, we again evaluate the imaginary part of @xmath6 from the available experimental data on the total cross section , while @xmath63 is calculated by the dispersion relations .
the details of the calculations and references to related dispersion analyses are given in ref .
we directly continue with the status of the data and proper parameterizations . fig .
[ elastic8 ] shows the experimental results @xcite on the @xmath3 and @xmath4 total cross sections together with the parameterizations from ref .
notice , that the data are well defined at @xmath59=800 mev / c .
the resulting ratio @xmath28 is shown in fig .
[ elastic8a ] as a function of the antikaon momentum for a proton and neutron target ; the @xmath3 experimental results are taken from refs .
@xcite .
it is important to note , that the experimental result on the real part of the forward scattering amplitude is determined from @xmath64 ^ 2 = \left .
\frac{d\sigma}{d\omega } \right|_{q=0 } - \,\left[\frac{k \sigma}{4 \pi}\right]^2,\ ] ] which does not provide the sign of @xmath5 .
however , the sign of @xmath63 can be fixed by the combined data analysis with the @xmath65-matrix and the dispersion relations @xcite by applying the crossing symmetry .
we also note , that the amount of experimental points shown in fig .
[ elastic8a ] for @xmath3 scattering is larger then that in the latest analysis by martin @xcite .
the situation for the real part of the @xmath4 forward scattering amplitude is different .
while @xmath66 can be reasonably determined from the low energy solution within the @xmath65-matrix analysis , there remains an uncertainty in the absolute normalization of the substraction point .
the latter one is fixed by the two experimental points @xcite shown in the lower part of fig .
[ elastic8a ] .
however , the experimental results do not indicate the sign of @xmath63 .
furthermore , there are no data available for the real part of the @xmath67 forward scattering amplitude . in principle
, the ratio @xmath5 for the @xmath4 scattering might stay positive and not change its sign around @xmath681.2 gev .
however , for these antikaon momenta the ratio @xmath5 is small and close to zero .
this important feature will be discussed later again . keeping in mind the uncertainties in the determination of the ratio @xmath5 for the @xmath4 interaction we now can perform a comparison with the data evaluated from the @xmath8 elastic scattering on @xmath12 and @xmath50 .
both , @xmath12 and @xmath50 , are isospin - symmetric nuclei and for a comparison with the values of @xmath26 and @xmath5 in free space we average over the results from the previous section for proton and neutron targets . in principle , the cross section @xmath26 and the ratio @xmath5 should be averaged over the fermi distribution available at densities @xmath690.1 @xmath62 , which is important for antikaon momenta where @xmath26 or @xmath5 rapidly change with @xmath59 .
however , this is not the case for @xmath59=800 mev / c .
the cross section @xmath26 and ratio @xmath5 taken from the free @xmath1 interactions averaged over equal numbers of protons and neutrons are shown in fig .
[ elastic8b ] by the solid lines .
the dots in fig .
[ elastic8b ] indicate the results evaluated from the data on @xmath8 elastic scattering ( see table [ tab1 ] ) averaged over the @xmath12 and @xmath50 targets .
we find that the total cross section extracted from nuclear elastic scattering is practically equal to the value given in free space , while the ratio @xmath5 differs by about a factor of @xmath141.5 .
when discussing this difference one should keep in mind that the calculations actually do not reproduce the data exactly at the diffractive minima leaving room for an uncertainty in the ratio @xmath5 , which is not well defined in free @xmath8-neutron interactions .
the factor @xmath141.5 might be attributed to the substraction point providing a global shift of @xmath5 ( cf .
[ elastic8a ] ) .
however , in this case the two experimental points @xcite for the real part of the @xmath4 forward scattering amplitude can not be reproduced .
we thus reject this possibility .
since the total @xmath1 cross section apparently is not modified in nuclear matter at densities @xmath690.1 @xmath62 and antikaon momenta of 800 mev / c relative to its value in free space , the drop in @xmath5 should be attributed to a decrease of the real part of the forward @xmath1 scattering amplitude .
adopting a spherically symmetric nuclear optical potential @xmath70 the phase shift @xmath23 is given as @xcite @xmath71 comparing to eq .
( [ shift1 ] ) this gives @xmath72 which can be rewritten in terms of @xmath26 and @xmath5 as @xmath73 the coulomb potential , furthermore , is given as @xcite @xmath74,\ ] ] where @xmath61 is the density distribution .
[ elastic9 ] shows the resulting real and imaginary part of the nuclear potential and the coulumb potential for @xmath8-meson scattering on @xmath12 and @xmath50 at an antikaon momentum of 800 mev / c .
the results for @xmath70 are shown for @xmath26 and @xmath5 as evaluated from the experimental data on the differential elastic cross section and given in table [ tab1 ] .
the coulomb potential is very small , however , important for the description of the differential data on the @xmath8 elastic scattering from nuclei .
the imaginary part of the optical potential clearly dominates and is close to that expected from free - space calculations .
the real part of the potential is attractive but smaller than expected from free - space calculations @xcite .
in this section we outline further perspectives of experiments on differential data for @xmath8 elastic scattering from nuclei .
there are two regions of antikaon momenta that are crucial for the understanding of the @xmath8 properties in nuclear matter and which can be studied by @xmath8-meson elastic scatterings from nuclei at densities @xmath690.1 @xmath62 . let us discuss both of them by starting with the high energy region . as shown in fig .
[ elastic8a ] the ratio @xmath5 of the real to imaginary part of the forward scattering amplitude in free @xmath8-neutron interactions is close to zero for antikaon momenta of @xmath141.2 gev / c . after averaging of @xmath5 over the number of protons and neutrons in the target the average @xmath75 0
the result for symmetric nuclei is shown in fig .
[ elastic8b ] .
we recall that neglecting coulomb scattering the calculations with @xmath5=0 yield a differential elastic cross section which is exactly equal to zero at the diffractive minima . taking into account the interference between the nuclear and coulomb phase shifts
the @xmath5-dependence of the cross section in the vicinity of the diffractive minima becomes more complicated @xcite , but can be predicted using @xmath26 and @xmath5 given in free space .
[ elastic2a ] shows the differential elastic cross section for @xmath8-meson scattering from @xmath12 at @xmath59=1.2 gev / c calculated with @xmath26=33.8 mb and @xmath5=0.02 , which corresponds to @xmath1 interactions in free space averaged over the number of protons and neutrons .
the calculations indicate a deep diffractive minimum due to the small value of the ratio @xmath5 .
we do expect that the measurements should allow to extract the actual value of @xmath5 and its sign .
this is important for the in - medium modification of the real part of the forward scattering amplitude @xmath6 as well as for a possible change of sign of @xmath63 around @xmath681.2 gev for the @xmath4 interaction .
a more sophisticated analysis can be done by scanning the antikaon momentum from 0.7 gev / c to 1.4 gev / c and a determination of the differential cross section at the first diffractive minimum , i.e. at its minimal point within the experimental resolution . as shown for the case of symmetric nuclear matter in fig .
[ elastic8b ] the cross section @xmath26 does not vary very much for antikaon momenta in this range while the ratio @xmath5 strongly depends on @xmath59 .
[ elastic2b ] shows the differential cross section at the first diffractive minimum for @xmath8 elastic scattering from @xmath12 in the antikaon momentum range [email protected] gev .
the solid line indicates the calculations performed with @xmath26 and @xmath5 taken from the free space @xmath1 scatterings .
the calculations are averaged at the diffractive minimum over the scattering angles with @xmath760.05 degrees .
the full dot in fig .
[ elastic2b ] shows the experimental result from ref .
@xcite while the result evaluated by the @xmath51 minimization is shown in terms of the full square at 0.8 gev / c .
since the predicted differential cross section at the diffractive minimum substantially depends on the antikaon momentum ( by two orders of magnitude ) we expect that the measurements should indicate a strong variation of @xmath77 even within the systematical uncertainties .
[ elastic8a ] indicates that in free space the real part of the forward scattering amplitude close to zero antikaon momenta is negative for @xmath3 and positive for @xmath4 interactions . by taking the free space
values the real part of the optical potential for isospin symmetric nuclei amounts to @xmath1418 mev at @xmath78 .
thus the real part of @xmath70 is attractive , but small , which contradicts the experimental data on kaonic atoms where a depth of the @xmath8 potential at @xmath10 of 180 mev @xcite has been extracted .
this descrepancy was resolved in the literature by considering the in - medium modification @xcite of the @xmath79 resonance which couples to the @xmath3 channel and yields a negative @xmath6 in free space at low antikaon momenta . the dominant effect , which modifies the @xmath79 dynamics in nuclear matter , stems from the pauli blocking of the intermediate nucleon states which causes a dissolution of this resonance @xcite and an in - medium modification of the @xmath3 forward scattering amplitude . for the most recent review of the problem as well as the current status on the analysis of kaonic atoms
the reader is refered to ref .
@xcite .
the contribution from the @xmath79 resonance to the total real part of the @xmath3 forward scattering amplitude is sizeable up to antikaon momenta of @xmath14400 mev / c @xcite .
thus the in - medium modification of @xmath6 might also be studied by @xmath8 elastic scattering from nuclei at low antikaon momenta @xmath80400 mev / c .
however , the crucial question here is the validity of the glauber theory at low energies . the limits of its applicability as well as the analysis of data on low energy antiproton elastic and inelastic scatterings from nuclei are given in ref .
it follows that glauber theory at low energies may serve only as a leading consideration .
[ elastic10 ] shows the differential cross section for @xmath8 elastic scattering from @xmath12 at an antikaon momentum of 300 mev / c .
the solid line indicates a calculation with cross section @xmath26 and ratio @xmath5 taken from the free space @xmath1 interactions while the dotted line shows the result obtained with the same @xmath26 but for @xmath5=0 . the dashed line in fig .
[ elastic10 ] shows the calculation with the modified real part of the forward @xmath3 scattering amplitude by discarding the contribution from the @xmath81(1405 ) resonance to @xmath63 . in the latter case
we obtain @xmath5=0.82 after averaging over protons and neutrons .
[ elastic10 ] indicates that the difference between the calculations with different ratios @xmath5 is sizeable .
although our predictions at @xmath59=300 mev / c may serve only as a leading extrapolation , we expect that the effect of the in - medium modification of the forward scattering amplitude can actually be measured by elastic @xmath0 scattering .
in this work we have considered the possibility to extract the antikaon potential from differential data on elastic @xmath0 scattering within the glauber theory and analyzed the experimental results @xcite for @xmath8 elastic scattering from @xmath12 and @xmath50 at an antikaon momentum of 800 mev / c .
we have found that the total @xmath1 cross section extracted from the data is close to the value given in free space @xmath3 and @xmath4 interactions when averaged over the number of protons and neutrons in the target . on the other hand , the ratio @xmath5 of the real to imaginary part of the forward scattering amplitude @xmath6 - as evaluated from the data - differs from the ratio @xmath5 in free space .
since the imaginary part of @xmath6 seems not to be modified this difference indicates an in - medium modification of the real part of @xmath6 .
however , the confidence level of this result ( within the @xmath51 method ) is below 10% which leaves it as an indication and does not allow for a final conclusion .
the parameters evaluated from @xmath8 scattering from @xmath12 and @xmath50 have been used for a reconstruction of the antikaon potential at @xmath59=800 mev / c , which is found to be attractive in line with the analysis from kiselev @xcite at 1.2 gev / c .
we have investigated the target nuclear densities that can be tested by @xmath0 elastic scattering and found them to range from 0 to @xmath61=0.1 @xmath62 . to extrapolate the depth of the @xmath8 potential at @xmath820.17 @xmath62
we have used nuclear density profiles with parameters evaluated from electron - nucleus scattering .
furthermore , we collect the information on the real part of the antikaon potential from all available sources for different antikaon momenta and show the results in fig .
[ elastic11 ] extrapolated to normal nuclear density @xmath61=0.17 @xmath62 .
the potential @xmath83 mev at @xmath59=0 stems from the data on kaonic atoms @xcite with the most recent analysis performed in ref .
the analyis of the experimental results @xcite on @xmath8-meson production from heavy - ion collisions is taken from refs .
@xcite and indicates @xmath84=80@xmath15120 mev at antikaon momenta @xmath85 mev / c .
[ elastic11 ] also shows the potential evaluated in this work from the data on @xmath54 and @xmath86 elastic scattering ; the indicated errorbar is due to a standard deviation corresponding to a 5% confidence level .
moreover , the solid line in fig .
[ elastic11 ] shows our result from ref .
@xcite calculated by discarding the contribution from @xmath81(1405 ) and @xmath87(1385 ) resonances at nuclear matter density .
note that the experimental results as well as the calculations indicate an attractive @xmath8 potential up to antikaon momenta of @xmath141.3 gev / c .
we , furthermore , have discussed future perspectives in @xmath0 elastic scatterings and the investigation of the antikaon potential or modification of the scattering amplitude @xmath6 .
it is pointed out that @xmath0 elastic scattering at low antikaon momenta may provide an additional ( obvious ) way to reconstruct the antikaon potential as an alternative to heavy - ion and proton- nucleus @xcite studies .
furthermore , measurements of the elastic @xmath8 scattering from nuclei at @xmath88 gev / c provide an effective tool to investigate , if the real part of the @xmath4 forward scattering amplitudy changes its sign around this momentum , and to learn more about the interaction between the antikaon and neutron .
kaplan and a.e .
nelson , phys .
b * 175 * , 57 ( 1986 ) g.e .
brown and m. rho , phys .
66 * , 2720 ( 1991 ) .
brown , phys .
bltter * 53 * , 671 ( 1997 ) .
v. thorsson , m. prakash and j.m .
lattimer , nucl .
a * 572 * , 693 ( 1994 ) ; a * 574 * , 851 ( 1994 ) .
j. schaffner , a. gal , i.n .
mishustin , h. stcker and w. greiner , phys .
, b * 334 * , 268 ( 1994 ) .
e. friedman , a. gal and g. batty , phys .
b * 308 * , 6 ( 1993 ) .
e. friedman , a. gal and g. batty , nucl .
phys . a * 579 * , 518 ( 1994 ) .
a. schrter et al . , z. phys .
a * 350 * , 101 ( 1994 ) .
p. senger et al . , acta phys .
b * 27 * , 2993 ( 1996 ) .
r. barth et al .
lett . * 78 * , 4007 ( 1997 ) .
f. laue et al . , phys .
lett . * 82 * , 1640 ( 1999 ) .
p. senger and h. strobele , j. phys .
g * 25 * , r59 ( 1999 ) .
li , c. m. ko and x.s .
fang , phys .
b * 329 * , 149 ( 1994 ) .
w. cassing et al .
a * 614 * , 415 ( 1997 ) .
lee and g.e .
brown , nucl .
a * 625 * , 372 ( 1997 ) .
e. l. bratkovskaya , w. cassing and u. mosel , nucl .
a * 622 * , 593 ( 1997 ) ; phys .
b * 424 * , 224 ( 1998 ) .
w. cassing and e. l. bratkovskaya , phys .
rep . * 308 * , 65 ( 1999 ) .
a. sibirtsev and w. cassing , nucl .
a * 641 * , 476 ( 1998 ) .
a. sibirtsev and w. cassing , nucl - th/9909024 .
glauber , _ lectures in theoretical physics _ , interscience publishers , new york ( 1959 ) p.315 .
newton , _ scattering theory of waves and particles _ , springer - verlag , berlin ( 1982 ) p.605 .
w. czyz , l.lesniak and h. woleck , nucl .
b * 19 * , 125 ( 1970 ) .
glauber and g. matthiae , nucl .
b * 21 * , 125 ( 1970 ) .
i. ahmad , nucl .
a * 247 * , 418 ( 1975 ) .
alkhasov , s.l .
belostotsky and a.a .
vorobyov , phys .
c * 42 * , 89 ( 1978 ) .
o.d . dalkarov and v.a .
karmanov , nucl . phys .
a * 445 * , 579 ( 1985 ) .
d. marlow et al .
rev c * 25 * , 2619 ( 1982 ) .
f. james and m. roos , comp .
. commun . * 10 * , 343 ( 1975 ) .
o.d . dalkarov and v.a .
karmanov , phys .
b * 147 * , 1 ( 1984 ) .
m. cresti , l. peruzzo and g. sartori , phys .
b * 132 * , 209 ( 1983 ) .
r. clapisch , nucl .
a * 434 * , 222 ( 1985 ) h. iwasaki et al .
b * 103 * , 247 ( 1981 ) .
kondratyuk and yu .
simonov , jetp let . * 17*,435 ( 1973 ) .
particle data group , eur .
j. c * 3 * , 1 ( 1998 ) .
b. conforto et al . , nucl .
b * 8 * , 263 ( 1968 ) .
r. armenteros et al .
b * 21 * , 13 ( 1970 ) .
dumbrais , t.yu .
dumbrais and n.m .
queen , fortschr .
* 19 * , 491 ( 1971 ) .
martin , nucl .
b * 179 * , 33 ( 1981 ) . v. koch , phys .
b * 337 * , 7 ( 1994 ) .
t. waas , m. rho and w. weise , nucl .
phys . a * 617 * , 449 ( 1997 ) .
a. ohnishi , y. nara and w. koch , phys .
c * 56 * , 2767 ( 1997 ) .
m. lutz , phys .
b * 426 * , 12 ( 1998 ) .
brown and m. rho , nucl .
a * 596 * , 503 ( 1996 ) .
t. waas , w. weise , nucl .
phys . a * 625 * , 287 ( 1997 ) .
e. friedman , a. gal , j. mares and a. ciepl , nucl - th/9804072 .
t. kiselev et al . ,
g * 25 * , 381 ( 1999 ) .
t. kirchner et al .
, cosy proposal 21 ; k. sistemich et al .
, strangeness in nuclei , krakow ( 1992 ) p.359 .
schult et al . ,
a * 583 * , 629c ( 1995 ) .
a. sibirtsev , h. mller and c. schneidereit , z. phys .
a * 351 * , 333 ( 1995 ) .
m. bscher et al .
, acta phys . pol .
* 27 * , 3087 ( 1996 ) . | we extract the antikaon potential from differential data on elastic @xmath0 scattering within glauber theory at an antikaon momentum of 800 mev / c .
the @xmath1 cross section at densities up to 0.1 @xmath2 is found to be close to the value given in free space @xmath3 and @xmath4 interactions ( averaged over the number of protons and neutrons ) whereas the ratio @xmath5 of the real to imaginary part of the forward scattering amplitude @xmath6 differs from the ratio @xmath5 in free space indicating a dropping of @xmath7 at finite nuclear density .
the extrapolated @xmath8 potential at a momentum of 0.8 gev / c amounts to @xmath928 mev at density @xmath10 which is substantially less than the values extracted from kaonic atoms at zero momentum or heavy - ion data at 300 mev / c@xmath11600 mev / c .
we , furthermore , investigate the perspectives of scattering experiments on @xmath12 at low ( @xmath13300 mev / c ) and high ( @xmath131.2 gev / c ) momenta . |
the issue of transversality in gromov s theory of pseudoholomorphic curves @xcite has always been problematic , and has attracted renewed interest in recent years . while many powerful symplectic invariants such as gromov - witten theory , hamiltonian floer homology and symplectic field theory are based on holomorphic curves , most of them run into severe technical complications unless multiply covered curves can be excluded , thus necessitating rather sophisticated techniques that typically replace the standard nonlinear cauchy - riemann equation by an abstract perturbation , see e.g. @xcite . aside from the technical challenges that these methods pose , they are non - ideal for many applications : for instance abstract perturbations destroy intersection theory in symplectic @xmath1-manifolds , and in calabi - yau @xmath2-folds they obscure information that one might hope to find in the geometric relationship between simple curves and their multiple covers , as exemplified by the gopakumar - vafa conjecture @xcite .
the motivating principle of this paper is in some sense orthogonal to that of abstract perturbations : our aim will be to extend the transversality theory for the _ standard _ pseudoholomorphic curve equation as far as it can reasonably be pushed , i.e. to prove transversality when it is possible , and in other cases to isolate the precise phenomena which make it impossible and explain what is true instead .
let us start by singling out two situations in which this program is not obviously hopeless .
[ ex : unbranched ] if @xmath3 is a closed @xmath4-holomorphic curve and @xmath5 is an _
unbranched _ cover of closed connected riemann surfaces with degree @xmath6 , then the virtual dimensions of the moduli spaces containing @xmath7 and @xmath8 , also known as the _ indices _ of these two curves , are related by @xmath9 since @xmath10 is then nonnegative whenever @xmath11 , there is no obvious reason why @xmath12 could not achieve transversality generically , but traditional methods in the theory of @xmath4-holomorphic curves do not prove this except when @xmath12 is simply covered , or in certain @xmath1-dimensional cases @xcite , or more recently , when @xmath13 if a sufficiently large space of perturbed almost complex structures is allowed @xcite .
[ ex : super ] suppose @xmath3 is a closed simply covered curve with index @xmath0 and @xmath5 is a branched cover of closed connected riemann surfaces with degree @xmath6 and @xmath14 as the algebraic count of branch points . then combining the riemann - hurwitz formula @xmath15 with the standard index formula for closed holomorphic curves gives the relation @xmath16 where @xmath17 .
this shows that @xmath12 can live in a space of negative virtual dimension when @xmath18 and thus can not generally achieve transversality , even though it will survive perturbations of @xmath4 as long as @xmath7 also does . it is interesting however to observe that @xmath7 must be immersed if @xmath4 is generic , so it has a well - defined normal bundle @xmath19 , and restricting the linearized cauchy - riemann operators for @xmath7 and @xmath12 to the normal bundle and its pullback gives operators @xmath20 and @xmath21 with indices related by @xmath22 the latter is always nonpositive , so @xmath23 can be injective , and this condition has a geometric meaning : as shown in proposition [ prop : isolated ] below , it implies that @xmath12 can never be the limit of a sequence of somewhere injective curves .
in fact , the only other curves near @xmath12 are other branched covers of the form @xmath24 for @xmath25 near @xmath26 , and the cokernels of the operators @xmath23 define an obstruction bundle over the space of branched covers which can be used to compute gromov - witten invariants .
this phenomenon is known as _ super - rigidity _ , see definition [ defn : super ] .
considerable interest in super - rigidity has been motivated by the study of gromov - witten invariants in calabi - yau @xmath2-folds , notably the gopakumar - vafa conjecture .
bryan and pandharipande observed in @xcite that while super - rigidity is a quite delicate question in the _ algebraic _ geometry of calabi - yau @xmath2-folds , it might reasonably be expected to hold generically in the symplectic setting .
theorem [ thm : super ] below states that this is true , in fact for generic @xmath4 , _ all _ somewhere injective closed @xmath4-holomorphic curves of index @xmath0 in symplectic manifolds of dimension greater than four ( and also curves of low genus in dimension four ) are super - rigid .
complementary to this , we will see in theorem [ thm : unbranched ] that everything one could hope for in the setting of example [ ex : unbranched ] is also true , and we will also be able to prove some transversality results for branched covers ( theorem [ thm : branched ] ) .
all of these results follow from a general picture of cauchy - riemann type operators with symmetries described in [ sec : thebigidea ] , which has its origins in taubes s work on the gromov invariant of symplectic @xmath1-manifolds @xcite .
while the results in this paper focus specifically on closed holomorphic curves , there is no obvious obstruction to applying the same techniques to study punctured curves in symplectic cobordisms , which we expect to have interesting applications in symplectic field theory @xcite and embedded contact homology @xcite .
a few special cases of super - rigidity in the punctured case have previously been observed in @xcite ; those examples were restricted to dimension four , but the results of the present article suggest that super - rigidity is likely to hold in considerably greater generality . to state the main results , assume @xmath27 is a symplectic manifold with @xmath28 and @xmath29 is a smooth almost complex structure that is compatible with @xmath30 , meaning that @xmath31 defines a riemannian metric on @xmath32 .
we fix also an open subset @xmath33 with compact closure , and consider the space @xmath34 of smooth @xmath30-compatible almost complex structures on @xmath32 that match @xmath29 outside of @xmath35 , with its natural @xmath36-topology .
note that the existence of a symplectic form on @xmath32 is not required for any of the arguments in this paper , but we are including it in the setup since it is important in applications all results could alternatively be stated and proved for the larger space of @xmath30-tame almost complex structures , or for arbitrary almost complex structures on a smooth ( not necessarily symplectic ) manifold . following the usual convention among symplectic topologists
, we will say that a subset of a topological space is a baire subset if it is comeager , i.e. it is a countable intersection of open and dense subsets .
the intersection of a countable sequence of baire subsets is again a baire subset , and by the baire category theorem , any baire subset of a complete metric space is dense .
we will say that a given property is true generically ( e.g. for generic @xmath4 ) whenever there exists a baire subset of the space of all admissible data ( e.g. in @xmath37 ) such that the property holds for all choices of data in that subset
. given @xmath38 , a closed connected riemann surface @xmath39 and a @xmath4-holomorphic curve @xmath3 , the index of @xmath7 is the integer @xmath40 where we abbreviate @xmath41 \rangle$ ] , @xmath42 : = u_*[\sigma ] \in h_2(m)$ ] .
a closed and connected @xmath4-holomorphic curve @xmath43 is said to be a ( @xmath44-fold ) multiple cover of @xmath7 if @xmath45 for some holomorphic map @xmath46 of degree @xmath47 , and @xmath7 is called simple if it is nonconstant and is not a multiple cover of any other curve .
the notion of super - rigidity was outlined already in example [ ex : super ] ; see definition [ defn : super ] for a more precise formulation .
we will also use the term _ fredholm regular _ to refer to the standard notion of transversality for moduli spaces of unparametrized @xmath4-holomorphic curves , cf .
proposition [ prop : regular ] below . in each of the following theorems ,
@xmath27 is a symplectic manifold of dimension @xmath48 with a compatible almost complex structure @xmath29 , and @xmath33 is an open subset with compact closure .
[ thm : super ] if @xmath49 , then there exists a baire subset @xmath50 such that for all @xmath51 , every simple @xmath4-holomorphic curve of index @xmath0 that intersects @xmath35 is super - rigid .
moreover , this result also holds when @xmath52 for all simple index @xmath0 curves of genus @xmath0 or @xmath53 .
super - rigidity has a number of well - known consequences , which are especially important in the case @xmath54 .
these are based partly on the observation that the space of all covers of super - rigid curves is an open and closed subset of the ambient moduli space of @xmath4-holomorphic curves we will prove the following statement in appendix [ sec : isolated ] , see also corollary [ cor : isolated ] .
[ prop : isolated ] suppose @xmath55 is a sequence of almost complex manifolds with @xmath56 in @xmath36 on some compact subset containing a super - rigid @xmath57-holomorphic curve @xmath58 .
then for sufficiently large @xmath59 , there exists a sequence of @xmath60-holomorphic curves @xmath61 with @xmath62 and @xmath63 in @xmath36 , and if @xmath64 is any sequence of smooth closed @xmath60-holomorphic curves gromov - convergent to a stable nodal @xmath57-holomorphic curve with image contained in @xmath65 , then for all @xmath59 sufficiently large , every @xmath64 is either a biholomorphic reparametrization or a multiple cover of @xmath66 .
applying gromov compactness and the standard implicit function theorem for simple curves , plus the fact that simple @xmath4-holomorphic curves are generically embedded and disjoint from each other in dimensions greater than four , this implies : [ cor : dimension6 ] for generic compatible @xmath4 in a closed symplectic @xmath67-manifold @xmath27 , there exist for each @xmath68 at most finitely many distinct simple @xmath4-holomorphic curves @xmath7 of genus @xmath69 with @xmath70 . using results of zinger @xcite and lee - parker @xcite ,
theorem [ thm : super ] also implies that the space of branched covers of an embedded index @xmath0 curve generically admits a well - defined obstruction bundle which can be used to compute gromov - witten invariants .
in particular , if @xmath49 and @xmath3 is an embedded @xmath4-holomorphic curve with index @xmath0 , one can apply @xcite*theorem 1.2 with no marked point constraints to study the space of @xmath4-holomorphic curves with image in @xmath71 , so that theorem [ thm : super ] establishes hypothesis ( b ) in zinger s result , implying that the cokernels of the normal operators @xmath72 for @xmath45 varying in the ( compactified ) space @xmath73 of @xmath44-fold branched covers of @xmath7 with arithmetic genus @xmath74 form a well - defined and oriented orbibundle @xmath75 with @xmath76 . here
@xmath77 again denotes the count of critical points of @xmath26 , which can be expressed in terms of @xmath44 , @xmath69 and @xmath74 via the riemann - hurwitz formula .
this is most interesting in the @xmath67-dimensional case , since @xmath78 then means that @xmath79 , and the count of solutions to an abstract perturbation of the holomorphic curve equation can then be computed as the euler number of this bundle .
recall that @xmath27 is called a symplectic calabi - yau @xmath2-fold if @xmath54 and @xmath80 . in this case corollary [ cor : dimension6 ] applies to _ all _ simple curves in @xmath32 , with the consequence that the gromov - witten invariants can be reduced to finite sums of _ local gromov - witten invariants _ for embedded curves , computed in terms of the euler classes of the obstruction bundles mentioned above .
this has well - known applications to the gopakumar - vafa conjecture , see @xcite .
we next state two results on transversality for multiple covers .
[ thm : unbranched ] there exists a baire subset @xmath50 such that for all @xmath51 , for every simple @xmath4-holomorphic curve @xmath3 intersecting @xmath35 and every unbranched holomorphic cover @xmath5 of closed riemann surfaces , the @xmath4-holomorphic curve @xmath8 is fredholm regular .
the case @xmath81 of theorem [ thm : unbranched ] has been proved previously in @xcite , though with stronger assumptions : for technical reasons , it was necessary in that paper to assume that @xmath71 is _ contained entirely _ in @xmath35 , and in dimension four also to allow perturbations of @xmath4 that are @xmath30-tame but not necessarily @xmath30-compatible .
the present paper uses a completely different approach to the transversality problem and is thus able to remove these restrictions .
it is generally harder to achieve transversality for covers @xmath12 with branch points , e.g. the index relation shows that @xmath82 can easily become negative in dimensions greater than six .
more seriously , if @xmath7 is regular , then one can always find a smooth family of other multiple covers near @xmath12 obtained by varying both @xmath7 and @xmath26 in their respective moduli spaces ; since the latter lives in a space of real dimension @xmath83 , the condition @xmath84 is evidently necessary in order for @xmath12 to be fredholm regular .
observe that if @xmath26 has @xmath85 critical values , then this condition is satisfied whenever @xmath86 : indeed , each critical value is the image of at most @xmath87 branch points ( counted algebraically ) , so we have @xmath88 and implies @xmath89 the next result states that the condition @xmath90 is also , in some sense , sufficient .
[ thm : branched ] there exists a baire subset @xmath91 such that the following holds for all @xmath51 .
suppose @xmath3 is a simple @xmath4-holomorphic curve intersecting @xmath35 and satisfying @xmath92 for some integer @xmath85 , and @xmath5 is a holomorphic branched cover of closed connected riemann surfaces with @xmath93 distinct critical values .
then there exists a @xmath4-holomorphic curve and a holomorphic branched cover @xmath94 such that @xmath95 , @xmath96 , @xmath97 and @xmath98 are arbitrarily @xmath36-close to @xmath7 , @xmath26 , @xmath99 and @xmath100 respectively , and @xmath101 is fredholm regular . note that whenever @xmath10 is _ strictly _ greater than @xmath102 , one can combine this result with the implicit function theorem to deduce the existence of simple @xmath4-holomorphic curves that are small perturbations of multiple covers of @xmath7 . as will be clear from the proof of theorem [ thm : branched ]
, one can also arrange for the perturbed branched cover @xmath96 to have the same number of critical points and critical values as @xmath26 , i.e. it is a nearby point in the same moduli space of holomorphic branched covers with fixed _ branching data _
[ sec : thebigidea ] .
the proofs of these theorems are inspired by the work of taubes @xcite , whose definition of the gromov invariant for symplectic @xmath1-manifolds required a special case of theorem [ thm : unbranched ] along with related bifurcation - theoretic results ( cf .
[ sec : bifurcations ] ) for multiply covered holomorphic tori . roughly speaking
, the idea is to study the local structure of spaces of the form @xmath103 where @xmath104 are fixed integers , @xmath7 varies in the moduli space of simple @xmath4-holomorphic curves and @xmath26 varies in the moduli space of holomorphic branched covers .
ideally , one would like to show that these spaces are smooth manifolds for generic @xmath4 , and to compute their codimensions in the space of pairs @xmath105 .
this turns theorems [ thm : super ] and [ thm : unbranched ] into `` dimension counting '' problems , as whenever one can show that the codimension of @xmath106 is larger than the dimension of the ambient space , one concludes that either @xmath107 or @xmath108 can not be nontrivial .
this discussion is oversimplified in at least three respects : first , we will not be able to find any nice structure on @xmath106 if @xmath26 varies in the space of _ all _ branched covers , but it will help to confine it to certain substrata of that space in which all branch points have prescribed branching orders . for similar reasons , it will also help to confine @xmath7 to substrata in which its number of critical points and their orders are constrained , and this is easily done . more seriously , the space @xmath106 as sketched above can have different codimensions on different components , as its codimension depends intricately on symmetry information which is ignored in .
we will therefore need to define a more elaborate version of @xmath106 which depends on a splitting of the operator @xmath72 into summands corresponding to irreducible representations of the ( generalized ) symmetry group of the cover .
this idea is borrowed directly from @xcite , though the details are somewhat more involved since , in contrast to the case of unbranched covers of tori , we can not assume that all covers are regular or that their symmetry groups are abelian .
we will see that once the formalism is developed in sufficient generality , it `` breaks the symmetry '' of @xmath72 enough to make dimension counting arguments much more effective .
a slightly different variation on the ideas in @xcite has been implemented by eftekhary to prove a partial result toward super - rigidity in dimension six , see @xcite .
here is an outline of the rest of the paper . after establishing some standard definitions and notation , [ sec : mainidea ]
will further elucidate the ideas sketched above and formulate a precise version of the statement that @xmath106 from is a smooth submanifold , theorem [ thm : submanifolds0 ] .
this will then be used as a black box to prove theorems [ thm : super ] , [ thm : unbranched ] and [ thm : branched ] in [ sec : everything ] , followed in [ sec : bifurcations ] by a brief informal discussion of bifurcation theory . the remainder of the paper is then devoted to the proof of theorem [ thm : submanifolds0 ] . in
[ sec : prep ] , we explain the splitting construction for cauchy - riemann operators with symmetries and prove some lemmas based on a mixture of elliptic regularity for punctured cauchy - riemann operators , topology of covering spaces , and representation theory of finite groups . the summands in the splitting
are also cauchy - riemann operators , whose indices are a somewhat delicate computation , carried out in [ sec : index ] . in
[ sec : unique ] we prove the special unique continuation result that is required for the sard - smale argument in [ sec : sardsmale ] , which completes the proof of theorem [ thm : submanifolds0 ] .
finally , [ sec : dimensionfour ] deals with super - rigidity in the four - dimensional case , which is something of an anomaly and requires different techniques based on intersection theory .
the two appendices prove a pair of results that are considered `` standard '' and yet , in this author s experience , seem sufficiently badly understood among experts to warrant some discussion ; the proofs require a few ideas that will in any case be useful elsewhere in the paper .
the super - rigidity problem has a slightly troubled history , and as the author of a new paper on the subject , it would behoove me at this point to apologize for having caused some of that trouble : i am aware of three previous attempts to prove some version of theorem [ thm : super ] which were later either withdrawn or revised to prove much weaker statements , and i was an author of one of them ( the original version of @xcite ) . with that in mind , i would sympathize with any reader s inclination to greet the present paper with a dose of skepticism , though it seems worth pointing out that rather than being an attempt to rescue the ( probably unrescuable ) proof originally attempted in @xcite , the approach taken in this paper has almost nothing in common with the previous one , other than the considerable debt that both of them owe to the ideas of taubes @xcite .
i would like to thank dan cristofaro - gardiner , chris gerig , michael hutchings , eleny ionel , tom parker , cliff taubes and aleksey zinger for conversations and correspondence which helped to improve my understanding of the problems studied in this paper .
let us now fix some notation and definitions that will be essential in the rest of the paper . given integers @xmath109 and a homology class @xmath110 , the moduli space of unparametrized @xmath4-holomorphic curves @xmath111 can be defined as the set of equivalence classes of tuples @xmath112 where @xmath39 is a closed connected riemann surface of genus @xmath69 , @xmath113 is an ordered set of distinct points ( the marked points ) , and @xmath3 is a @xmath4-holomorphic map satisfying @xmath42 : = u_*[\sigma ] = a$ ] , with equivalence defined by @xmath114 for diffeomorphisms @xmath115 .
the gromov compactification of @xmath111 is the space @xmath116 of ( equivalence classes of ) stable nodal curves @xmath117 , where now @xmath118 may be disconnected , and the original data are augmented by an unordered set of distinct points in @xmath119 , arranged into unordered pairs @xmath120 such that @xmath121 for each @xmath122 .
we call the pairs @xmath123 nodes , and each individual @xmath124 or @xmath125 a nodal point .
the curves in @xmath116 are required to have arithmetic genus @xmath69 , which means that the surface obtained from @xmath118 by performing connected sums at all matched pairs of nodal points is a closed connected surface of genus @xmath69 .
the stability condition requires that any component of @xmath126 on which @xmath7 is constant should have negative euler characteristic . with this condition , @xmath116 can be given a natural topology as a metrizable hausdorff space , and it is compact whenever @xmath4 is tamed by a symplectic form .
a definition of the topology may be found e.g. in @xcite ; for sequences in @xmath111 , it amounts to the notion of @xmath36-convergence for @xmath99 and @xmath7 after a choice of parametrization for which all domains and marked point sets are identified .
curves @xmath127 \in \overline{{{\mathcal m}}}_{g , m}(a , j)$ ] with @xmath128 can equivalently be regarded as elements of @xmath111 , and are thus called smooth curves to distinguish them from nodal curves . in this paper ,
the word `` curve '' always means `` smooth curve '' ( i.e. without nodes ) unless the word `` nodal '' is explicitly included .
similarly , all dimensions and fredholm indices in this paper are _ real _ ( not complex ) unless otherwise specified .
this usage differs somewhat from the algebraic geometry literature .
when there is no danger of confusion , we shall sometimes abuse notation by denoting equivalence classes @xmath129 \in { { \mathcal m}}_{g , m}(a , j)$ ] or @xmath130 \in \overline{{{\mathcal m}}}_{g , m}(a , j)$ ] simply by @xmath131 or @xmath132 respectively , and we will refer to the restriction of a nodal curve @xmath127 $ ] to any connected component of its domain @xmath118 as a smooth component of @xmath7
. we shall also abbreviate @xmath133 recall that @xmath134 has virtual dimension equal to the index of any curve @xmath135 , while the virtual dimension of the moduli space with marked points is @xmath136 the multiply covered curves form a distinguished closed subset of @xmath137 .
given any @xmath135 with domain @xmath39 , and integers @xmath138 , @xmath139 , define the space of stable nodal @xmath44-fold covers of @xmath7 , @xmath140 \in \overline{{{\mathcal m}}}_{h}(da , j)\ \big|\ [ ( s,\tilde{\jmath},\delta,\varphi ) ] \in \overline{{{\mathcal m}}}_{h}(d[\sigma],j ) \right\},\ ] ] so in particular , each smooth component @xmath141 of @xmath142 belongs to a space @xmath143 of smooth branched covers @xmath144 of some degreee @xmath145 , such that @xmath146 . note that @xmath73 may in general be strictly larger than the closure of @xmath147 in the gromov topology to cite one well - known example , the space @xmath148,i)$ ] of smooth degree @xmath53 holomorphic tori in @xmath149 is empty , but @xmath150,i)$ ] contains a nodal curve with a constant component of genus @xmath53 . in this language ,
the main consequence of proposition [ prop : isolated ] can be stated as follows .
[ cor : isolated ] suppose @xmath151 is an almost complex manifold and @xmath152 is a super - rigid curve in @xmath32 . then for every @xmath138 and @xmath139 , @xmath153 is an open and closed subset of @xmath154 .
recall next that every @xmath4-holomorphic curve @xmath3 gives rise to a linearized cauchy - riemann operator @xmath155 i.e. the linearization at @xmath7 of the nonlinear cauchy - riemann operator @xmath156 , whose zero - set is the space of all @xmath4-holomorphic maps .
the operator @xmath157 takes vector fields along @xmath7 to @xmath158-forms valued in the complex vector bundle @xmath159 , and can be written explicitly as @xmath160 for any choice of symmetric connection @xmath161 ( cf .
@xcite*2.4 ) .
recall moreover that whenever @xmath7 is nonconstant , its critical points are isolated and one can find a smooth splitting of complex vector bundles @xmath162 such that @xmath163 matches the image of @xmath164 at regular points ; see e.g. @xcite*3.3 for details . in many cases of interest in this paper
, @xmath7 will be a cover of an immersed @xmath4-holomorphic curve @xmath165 , so @xmath166 is then simply the pullback of the normal bundle of @xmath165 via the cover .
we define the normal cauchy - riemann operator at @xmath7 as the restriction of @xmath157 to sections of @xmath166 , composed with the projection @xmath167 , hence @xmath168 in general , a neighborhood of any element in @xmath111 can be identified with the zero - set of a smooth fredholm section of a banach space bundle , modulo a finite group action if there are nontrivial automorphisms .
we say that @xmath135 is fredholm regular whenever it is a transverse intersection of this section with the zero - section .
note that whenever this condition holds , it automatically also holds after adding any finite collection of marked points and viewing @xmath7 as an element of @xmath111 .
the implicit function theorem gives the open set of regular curves in @xmath111 the structure of a smooth orbifold with dimension equal to its virtual dimension , and local isotropy groups determined by the automorphism groups of the curves in particular , the set of regular simple curves forms a manifold , though orbifold singularities can appear when multiple covers are included .
the following convenient repackaging of the regularity condition comes from @xcite*corollary 3.13 .
[ prop : regular ] a closed and connected @xmath4-holomorphic curve @xmath3 is fredholm regular if and only if its normal cauchy - riemann operator @xmath169 is surjective for some ( and therefore all ) @xmath170 and @xmath171 .
[ defn : super ] a closed , connected , simple @xmath4-holomorphic curve @xmath3 is called super - rigid if it satisfies the following : 1 .
@xmath13 ; 2 .
@xmath172 is an immersion ; 3 . for all closed connected riemann surfaces @xmath173 and holomorphic maps @xmath174 of positive degree
, the curve @xmath175 admits no nontrivial solutions to the normal linearized equation @xmath176 .
we now explain in precise terms the stratification result that underlies the main theorems of [ sec : maintheorems ] .
suppose @xmath177 is a simple curve with genus @xmath68 and @xmath178 is a branched cover with degree @xmath179 , giving rise to the multiply covered curve @xmath180 of genus @xmath138 .
for the sake of intuition , we begin under the assumption @xmath181 .
then the automorphism group @xmath182 contains a unique nontrivial element @xmath183 , and the space of sections @xmath184 has a natural splitting @xmath185 where @xmath186 . splitting @xmath187 and suitable sobolev spaces of sections on these two bundles in the same way ,
one obtains a splitting of the normal cauchy - riemann operator @xmath188 into two operators @xmath189 .
it is not hard to see that @xmath190 is in some sense equivalent to @xmath191 , as its domain and target both consist of sections that are pullbacks via @xmath26 of sections over @xmath192 .
the standard transversality theory for simple curves thus shows that @xmath193 can be assumed surjective ( and also injective if @xmath165 is immersed with index @xmath0 ) if @xmath4 is chosen generically .
we will see that the problem of proving surjectivity or injectivity for @xmath20 becomes tractable when viewed as two independent problems for the operators @xmath190 and @xmath194 . in order to generalize this discussion beyond the degree @xmath195 case
, it helps to adopt an alternative perspective based on representation theory .
let @xmath113 denote a finite subset that contains all critical values of @xmath26 , and set @xmath196 so that @xmath197 is a smooth covering map with @xmath198 as its group of deck transformations .
define @xmath199 as the isomorphism to the symmetric group on @xmath200 .
we can then identify the covering map @xmath197 with @xmath201 \mapsto \varphi(z),\ ] ] where @xmath202 acts on @xmath203 by deck transformations and on @xmath200 via @xmath204 .
now if @xmath205 denotes the standard basis of @xmath206 , then @xmath204 also gives rise to a real permutation representation @xmath207 and a corresponding real vector bundle @xmath208 defined as the @xmath209-quotient of a trivial bundle over @xmath203 , @xmath210 the space of sections of the twisted normal bundle @xmath211 then has a natural identification with the space of sections of @xmath212 : indeed , we can represent sections of @xmath213 as @xmath209-equivariant sections @xmath214 of @xmath215 , which satisfy the relation @xmath216 , thus a corresponding section @xmath217 can be defined under the identification of @xmath218 with @xmath219 by @xmath220 ) = \eta^i(z).\ ] ] under this identification , @xmath20 becomes a cauchy - riemann type operator on the twisted bundle @xmath213 , defined locally by @xmath221 whenever @xmath165 is a local section of @xmath222 that has a constant lift to the trivial bundle @xmath223 .
the above construction appears cumbersome at first glance , but it has the following advantage : the decomposition @xmath224 now corresponds to a splitting of the twisted bundle @xmath213 into subbundles @xmath225 where @xmath226 are defined in terms of the natural splitting of @xmath227 into irreducible @xmath202-invariant subspaces @xmath228 this is the simplest nontrivial example of what turns out to be a general principle : splittings of cauchy - riemann operators for multiply covered curves arise from decompositions of permutation representations into irreducible summands . to turn @xmath229 into a splitting of cauchy - riemann operators
, we still have a small analytical issue to cope with since the bundles @xmath230 are defined over @xmath218 and do not both extend over the punctures . in place of , we therefore obtain a splitting @xmath231 where the dots over the operators indicate that we are restricting them to the punctured domain @xmath203 .
we will see in [ sec : crpunctured ] how to define suitable weighted sobolev spaces over @xmath218 and @xmath203 so that the punctured operators have the same indices , kernels and cokernels as their unpunctured counterparts .
we return now to the general case and continue using the notation @xmath197 for the covering map obtained by deleting some finite subsets that include the critical values and their preimages .
recall that @xmath26 is called regular if @xmath232 .
this condition was secretly important in the above discussion of the @xmath181 case , as the definition of the twisted bundle @xmath213 required identifying @xmath218 with the quotient of @xmath203 by deck transformations .
in general , @xmath233 may be trivial , but we can use some notions from elementary covering space theory to get around this .
[ defn : genaut ] the generalized automorphism group of a @xmath44-fold branched cover @xmath234 is the quotient @xmath235 , where @xmath236 is the normal core of @xmath237 , and @xmath218 and @xmath203 are defined by with @xmath238 as the set of critical values of @xmath26 .
[ defn : presentation ] a regular presentation of the holomorphic @xmath44-fold branched cover @xmath178 is a tuple @xmath239 consisting of : * a finite subset @xmath113 containing the critical values of @xmath26 and defining the punctured surfaces @xmath218 and @xmath203 via ; * a connected surface @xmath240 and regular covering map @xmath241 with finite automorphism group @xmath242 ; * a set @xmath243 with @xmath44 elements ; * a transitive action of @xmath202 on @xmath243 , defined via a homomorphism @xmath244 from @xmath202 to the symmetric group on @xmath243 ; * a diffeomorphism @xmath245 , where @xmath202 acts on @xmath240 by deck transformations and on @xmath243 via @xmath204 , such that @xmath246 takes the form @xmath247 \mapsto \pi(z).\ ] ] we say that @xmath239 is minimal if @xmath113 is the set of critical values of @xmath26 and @xmath244 is injective .
two regular presentations @xmath248 of @xmath234 for @xmath249 are isomorphic if @xmath250 and there exists a diffeomorphism @xmath251 , a bijection @xmath252 , and a group isomorphism @xmath253 such that : 1 .
@xmath254 and for all @xmath255 , @xmath256 ; 2 . for all @xmath255 , @xmath257 ; 3 .
@xmath258 takes the form @xmath259 \mapsto [ ( \psi(z),\beta(i))].\ ] ] we will show in [ sec : regpres ] that the regular cover @xmath260 in any regular presentation can be extended to a holomorphic branched cover of closed connected riemann surfaces @xmath261 such that @xmath262 .
observe that if @xmath263 and @xmath264 denotes the stabilizer of @xmath265 under the @xmath202-action defined by @xmath204 , then @xmath266 \mapsto [ ( z , i)]\ ] ] is a diffeomorphism identifying @xmath246 with the natural projection @xmath267 .
thus one can associate to any regular presentation a ( non - unique ) factorization of @xmath260 by covering maps @xmath268 , which extends over the punctures to a factorization of @xmath269 by holomorphic branched covers @xmath270 we will also show in lemma [ lemma : regpres ] that @xmath234 always admits a unique isomorphism class of minimal regular presentations @xmath239 , for which @xmath202 is isomorphic to the generalized automorphism group of @xmath26 , and in this case @xmath260 is isomorphic to @xmath271 whenever the latter happens to be already regular . given a choice of regular presentation @xmath239 , the discussion of the degree @xmath195 case can be generalized as follows .
the transitive action @xmath244 induces a permutation representation @xmath272 on the real vector space @xmath273 with basis labeled by the elements of @xmath243 , and a twisted bundle @xmath274 , where @xmath275 with a natural isomorphism @xmath276 that identifies @xmath20 with a cauchy - riemann operator @xmath277 on suitable exponentially weighted sobolev spaces of sections of @xmath213 .
any finite - dimensional real representation @xmath278 similarly gives rise to a twisted bundle @xmath279 , with @xmath280 , and a twisted cauchy - riemann operator @xmath281 which ( up to conjugacy ) depends only on @xmath282 and the isomorphism classes of the regular presentation and the representation @xmath283 .
now any representation - theoretic decomposition @xmath284 induces a splitting of the punctured cauchy - riemann operator @xmath285 with the following useful property : [ lemma : splitting ] the normal cauchy - riemann operator @xmath20 for a multiple cover is surjective or injective if and only if the same holds for all of the summands @xmath286 in .
we will see below that the splitting can be arranged to vary smoothly as @xmath165 and @xmath26 move about in their respective moduli spaces , so the indices of the summands @xmath286 are constant .
this immediately gives rise to `` no - go '' results about transversality and super - rigidity : the former is impossible unless all the @xmath286 have nonnegative index , and the latter unless they have nonpositive index .
conversely , whenever either of these index conditions holds for all summands given by irreducible representations , theorem [ thm : submanifolds0 ] below will imply that the desired transversality or super - rigidity result holds for all pairs @xmath287 lying in some open and dense subset .
this is the main idea behind theorem [ thm : branched ] , and it similarly provides a necessary and sufficient criterion for the feasibility of obstruction bundle arguments . it should be emphasized that the representations of @xmath202 in this discussion are _ real _ , not complex .
we will need to use the standard fact ( see [ sec : representations ] ) that for any finite group @xmath202 , real irreducible representations @xmath278 come in three types , characterized via the space @xmath288 of @xmath202-equivariant real - linear maps @xmath289 : * real type : @xmath290 * complex type : @xmath291 * quaternionic type : @xmath292 in order to discuss what happens to as @xmath165 and @xmath26 move in their respective moduli spaces , we observe that the construction depends quite heavily on the branching structure of @xmath234 , i.e. the number of punctures @xmath293 and the topological behavior of @xmath26 in their vicinity .
this necessitates decomposing the space of all degree @xmath44 branched covers into strata @xmath294,j ) = \bigcup_{\mathbf{b } } { { \mathcal m}}^d_{\mathbf{b}}(j)\ ] ] labeled by their so - called _ branching data _
choose an integer @xmath85 , and associate to each of the numbers @xmath122 a nonempty finite ordered set of natural numbers @xmath296 such that @xmath297 and at least one of the numbers @xmath298 is strictly greater than @xmath53 .
we denote the totality of this data by @xmath299 and call it branching data of degree @xmath44 with @xmath93 critical values . given this ,
let @xmath300 denote the moduli space of all closed and connected unparametrized @xmath99-holomorphic curves @xmath26 in @xmath39 of degree @xmath44 with @xmath301 marked points @xmath302 such that 1 .
there are distinct points @xmath303 such that @xmath304 for each @xmath122 ; 2 . for each @xmath122 and @xmath305
, @xmath26 is locally @xmath306-to-@xmath53 near @xmath307 ; and 3 .
@xmath26 has no critical points outside of the marked points .
note that we do not require _ every _ marked point of @xmath26 to be a critical point , but we are assuming @xmath308 is the set of critical values , whose preimages are marked points and may include both critical and regular points . for any @xmath309 , we have @xmath310 thus @xmath44 and @xmath295 determine the genus @xmath74 of @xmath26 via the riemann - hurwitz formula , and we shall denote by @xmath311,j)\ ] ] the image of the natural map @xmath312,j)$ ] defined by forgetting the marked points .
note that in some cases , the riemann - hurwitz calculation may produce a negative genus , which just means that @xmath313 is empty .
if @xmath295 is empty , i.e. @xmath314 , it means every @xmath315 is unbranched .
it is a classical fact that @xmath313 is a smooth manifold of real dimension @xmath316 , as it can be parametrized locally by the positions of the critical values @xmath303 .
moreover , it depends smoothly on @xmath99 in the sense that if @xmath317 is any smooth finite - dimensional family of complex structures on @xmath192 , then @xmath318 defines a smooth fiber bundle .
we will show in [ sec : regpres ] that regular presentations of @xmath234 can also be arranged to vary smoothly as @xmath26 varies with fixed branching data .
we are now nearly ready to state the main technical result of the paper .
given integers @xmath319 and @xmath320 , let @xmath321 denote the subset consisting of curves that have critical points of critical order @xmath322 at the @xmath265th marked point for @xmath323 and are immersed everywhere else .
as explained in appendix [ sec : critical ] , the simple curves in this space form a smooth submanifold for generic @xmath4 , with codimension @xmath324 in @xmath111 . given a finite group @xmath202 , an integer @xmath6 and branching data @xmath295 of degree @xmath44 with @xmath85 critical values , define @xmath325 so be the set of all curves with representatives of the form @xmath326 where @xmath178 represents an element in @xmath327 with generalized automorphism group isomorphic to @xmath202 , and @xmath177 is a simple curve with @xmath328 marked points representing an element of @xmath329 that intersects the subset @xmath33 .
note that if @xmath4 is generic on @xmath35 , then standard results give @xmath330 the structure of a smooth manifold with @xmath331 now fix a complete list of pairwise non - isomorphic irreducible real representations @xmath332 and for any @xmath333-tuples of nonnegative integers @xmath334 and @xmath335 , let @xmath336 denote the subset determined by the conditions @xmath337 where the twisted operators @xmath338 are defined with respect to the unique isomorphism class of minimal regular presentations . [ thm : submanifolds0 ] there exists a baire subset @xmath50 such that the following holds for all @xmath51 .
for all choices of a finite group @xmath202 , branching data @xmath295 , integers @xmath339 , @xmath320 and a homology class @xmath110 , the subset @xmath336 is a smooth submanifold with @xmath340 where for each @xmath341 , @xmath342 according to whether the representation @xmath343 is of real , complex or quaternionic type respectively .
[ remark : bifurcation ] the statement above is specifically geared toward the applications treated in this paper , but for different purposes one could formulate various other versions , e.g. one could add more marked points to @xmath329 and impose intersection constraints on them , or one could consider generic finite - dimensional families @xmath344 of almost complex structures and thus replace @xmath329 with a parametric moduli space of pairs @xmath345 where @xmath346 and @xmath7 is @xmath347-holomorphic .
either would require no serious modifications to the proof , other than more cumbersome notation .
we need two further ingredients in order to turn theorem [ thm : submanifolds0 ] into a powerful enough tool for proving the theorems of [ sec : maintheorems ] .
the first is an index calculation for the twisted operators @xmath281 .
the precise result is stated and proved in [ sec : index ] , but for the main applications we only need the following estimate , which is a corollary : [ lemma : index ] given a @xmath4-holomorphic curve @xmath177 with normal cauchy - riemann operator @xmath191 , a @xmath44-fold branched cover @xmath178 with @xmath85 critical values , a regular presentation @xmath239 for @xmath26 and a real representation @xmath278 , the resulting twisted cauchy - riemann operator @xmath281 for @xmath348 satisfies @xmath349 \le { \operatorname{ind}}(\dot{\mathbf{d}}_{u,\boldsymbol{\theta}}^n ) \le \dim w \cdot { \operatorname{ind}}(\mathbf{d}_v^n).\ ] ] moreover , if the regular presentation is minimal and @xmath283 is faithful , then the second estimate can be improved to @xmath350 and it is strict unless all branch points of @xmath26 have branching order @xmath195 . for the proof of super - rigidity , we will need the next result as a means of improving the upper bound in lemma [ lemma : index ] for representations that are not faithful .
[ lemma : nonfaithful ] under the assumptions of lemma [ lemma : index ] , suppose the regular presentation is minimal , and the splitting of @xmath20 includes a summand @xmath281 for which the representation @xmath278 is not faithful . then @xmath178 admits a factorization by holomorphic branched covers @xmath351 with @xmath352 , and @xmath281 is conjugate to an operator @xmath353 defined with respect to a regular presentation @xmath354 for @xmath355 , where @xmath356 , @xmath357 , and @xmath358 is the faithful representation of @xmath359 determined by @xmath283 .
moreover , @xmath360 also admits a splitting in the form which has @xmath353 as a summand .
let us now take the results of the previous section as black boxes and prove the main theorems from [ sec : maintheorems ] .
we argue by induction on the degrees @xmath6 of branched covers . for @xmath361
, we only need to know that generic perturbations of @xmath4 suffice to make all simple index @xmath0 curves through @xmath35 regular and immersed ; this is standard ( see appendix [ sec : critical ] for the immersion property ) .
thus for @xmath47 , assume we have already found a baire subset in @xmath37 for which all branched covers @xmath362 with @xmath177 a simple curve of index @xmath0 and @xmath363 have @xmath20 injective .
suppose @xmath364 has @xmath85 critical values and @xmath365 and @xmath20 is not injective for @xmath362 . then picking the minimal regular presentation @xmath239 for @xmath26 and decomposing @xmath366 into irreducible representations @xmath367 of @xmath202 splits @xmath20 into twisted cauchy - riemann operators @xmath338 for @xmath341 with @xmath368 and at least one of the @xmath369 must be strictly positive by lemma [ lemma : splitting ] . if @xmath370 and @xmath343 is non - faithful , then lemma [ lemma : nonfaithful ] identifies @xmath338 with a summand of @xmath360 for some other cover @xmath371 of @xmath165 with strictly smaller degree , implying @xmath372 and thus violating the inductive hypothesis .
we can therefore assume @xmath370 for some faithful representation @xmath343 .
but then theorem [ thm : submanifolds0 ] and lemma [ lemma : index ] imply that @xmath7 lives in a submanifold of the @xmath316-dimensional space of branched covers of @xmath165 with branching data @xmath295 , having dimension at most @xmath373 \le 2r - t_i k_i[k_i + ( n-1)r ] = r [ 2 - t_i k_i ( n-1 ) ] - t_i k_i^2 < 0\ ] ] since we are assuming @xmath374 .
this gives a contradiction and thus completes the induction . in dimension four
, the above argument fails to exclude the possibility of @xmath375 for some real - type repesentation @xmath343 , and this is why we do not know whether super - rigidity always holds in dimension four .
we will prove in [ sec : dimensionfour ] that it does hold for covers of genus zero and one curves , using different techniques based on intersection theory .
suppose @xmath177 is a simple curve intersecting @xmath35 and @xmath178 is a @xmath44-fold unbranched cover for which @xmath362 is not regular , hence by prop .
[ prop : regular ] , @xmath20 is not surjective .
fixing the minimal regular presentation of @xmath26 and considering the splitting , we find a twisted cauchy - riemann operator @xmath338 with @xmath376 for some irreducible representation @xmath343 of the generalized automorphism group @xmath202 of @xmath26 .
suppose @xmath165 has exactly @xmath319 critical points , with critical orders @xmath377 , so viewing these as marked points allows us to consider @xmath165 as an element in the space @xmath329 , which has dimension @xmath378 the count of critical points @xmath379 also appears in the relation between @xmath380 and @xmath381 : indeed , writing @xmath382 , we can view @xmath383 as a holomorphic section of @xmath384 , hence @xmath385 implying @xmath386 . plugging in this into the riemann - roch formula then gives @xmath387 meanwhile , @xmath26 lives in the discrete stratum of the space of branched covers since it has no branch points , and lemma [ lemma : index ] reduces to an equality @xmath388 now using theorem [ thm : submanifolds0 ] , we find that if @xmath4 is generic , @xmath7 lives in a manifold of dimension at most @xmath389 ) \\ & = ( 1 - t_i c_i \dim w ) \left [ { \operatorname{ind}}(v ) + 2 m - 2 n z(dv)\right ] \\ & \qquad - 2 t_i c_i \dim w \cdot \left [ ( n-1 ) z(dv ) - m \right ] - t_i c_i^2 < 0 , \end{split}\ ] ] where we note that @xmath390 since @xmath391 and every critical point has order at least @xmath53 .
assume @xmath177 is simple and satisfies @xmath392 , while @xmath178 has degree @xmath6 and @xmath93 critical values .
if @xmath4 is generic , then by proposition [ prop : critical ] the moduli space containing @xmath165 has an open and dense subset consisting of immersed curves , so we are free to assume @xmath165 is immersed and thus @xmath393 . the key observation is then that by lemma [ lemma : index ] , the twisted operators @xmath281 all have nonnegative index , hence theorem [ thm : submanifolds0 ] implies that all of them are surjective unless @xmath287 lies in a countable union of submanifolds with positive codimension .
part of the point of taubes s twisted bundle setup in @xcite was to understand bifurcations of isolated @xmath4-holomorphic tori under generic @xmath53-parameter deformations in @xmath4 . while bifurcation theory is not the main topic of this article , it should be clear that such a theory could be developed based on theorem [ thm : submanifolds0 ] , thus we take this opportunity to make a few observations about it .
if @xmath394}$ ] is a generic homotopy of compatible almost complex structures whose endpoints are generic , then as mentioned in remark [ remark : bifurcation ] , one can modify theorem [ thm : submanifolds0 ] to the statement that the parametric moduli space @xmath395 consisting of pairs @xmath345 where @xmath396 $ ] and @xmath397 contains certain smooth submanifolds characterized by the dimensions of the kernels and cokernels of twisted cauchy - riemann operators , and their codimensions are given by the same formula . in this
setting , suppose @xmath398 is a smooth @xmath53-parameter family of simple @xmath399-holomorphic curves with index @xmath0 for some function @xmath400 $ ] , and @xmath401 defines a corresponding @xmath53-parameter family of unbranched covers .
the latter have index @xmath0 and will be regular for almost every @xmath402 , but a bifurcation or `` wall crossing '' phenomenon occurs at any parameter value @xmath403 for which the family @xmath404 passes ( necessarily transversely ) through one of the codimension @xmath53 strata given by theorem [ thm : submanifolds0 ] . when this happens , most of the twisted operators @xmath405 remain both injective and surjective , but there will be exactly one irreducible representation @xmath283 for which @xmath406 and @xmath283 is necessarily of real type . whenever @xmath283 is not faithful
, one can factor @xmath407 through a cover @xmath408 of smaller degree and instead examine @xmath409 , so that @xmath283 becomes faithful without loss of generality ( cf .
lemma [ lemma : nonfaithful ] ) . for the trivial representation , this means replacing @xmath410 with @xmath411 itself , so regularity fails for the underlying simple curve at @xmath412 : as shown in @xcite , this is the case where the family @xmath398 undergoes a _ birth - death _ bifurcation .
the other interesting phenomenon examined by taubes was the _ degree - doubling _ bifurcation , in which @xmath411 remains regular but it has a double cover @xmath413 which loses regularity at @xmath412 , causing an additional @xmath53-parameter family of simple curves @xmath414 to collide with @xmath404 at @xmath412 .
this is what happens when @xmath415 remains an isomorphism for the trivial representation but acquires @xmath53-dimensional kernel and cokernel for the nontrivial irreducible representation of @xmath209 . in @xcite ,
no further bifurcations beyond these two types are possible : this can be attributed to the fact that since taubes only considers unbranched covers of tori , all covers are regular and abelian . as a consequence ,
all the complex irreducible representations in the picture are @xmath53-dimensional , implying that the only faithful real - type irreducible representations one needs to consider are the trivial representation of the trivial group and the nontrivial representation of @xmath209 .
we should not expect this fortunate situation to hold more generally : for unbranched covers with higher genus , one certainly encounters generalized automorphism groups that are non - abelian and thus have faithful real - type representations of dimension greater than one
. these should presumably give rise to bifurcation phenomena involving covers of arbitrarily high degree .
in the context of super - rigidity , it is also important to consider bifurcations that involve branched covers of index @xmath0 curves under generic homotopies of @xmath4 . inspecting the proof of theorem [ thm : super ]
, one should expect to see interesting phenomena whenever the dimension that was estimated at the end of the proof turns out to be at least @xmath416 , i.e. @xmath417 \ge -1.\ ] ] assuming we re in dimension at least six , this can only mean @xmath418 and either @xmath314 or @xmath78 .
the case @xmath314 means the cover is unbranched , so this is what we discussed in the previous paragraphs .
bifurcations involving branched covers can evidently also occur in dimension six , and in this case the improved index bound from lemma [ lemma : index ] must be an equality .
the scenario is therefore that the rank of the obstruction bundle over the space of covers @xmath419 jumps up by @xmath53 at a particular parameter value @xmath412 and for some isolated element @xmath420 in the space of branched covers with only simple ( i.e. two - to - one ) branch points : this can presumably cause both a change in the euler number of the obstruction bundle and the breaking off of a new family of simple curves from @xmath421 .
once again the irreducible representation involved must be of real type but can have arbitrary dimension , meaning we should not expect any limitation on the degree of @xmath420 , contrary to the situation in @xcite .
in this section we give a detailed account of the twisted bundle formalism behind theorem [ thm : submanifolds0 ] and prove several lemmas required for its proof , as well as lemma [ lemma : nonfaithful ] . instead of talking directly about @xmath4-holomorphic curves , we shall work in the context of abstract cauchy - riemann operators on vector bundles and their pullbacks .
the following lemma allows us to move freely back and forth between talking about holomorphic branched covers of closed riemann surfaces and honest covering maps of punctured surfaces .
[ lemma : closedup ] suppose @xmath422 is the complement of a finite set of points @xmath238 in a closed connected riemann surface @xmath39 , @xmath423 is a connected noncompact riemann surface , and @xmath424 is a holomorphic covering map of finite degree
. then there exists a closed connected riemann surface @xmath425 with a finite set of points @xmath426 such that @xmath423 admits a biholomorphic identification with @xmath427 and @xmath26 extends over the punctures to a holomorphic branched cover @xmath178 with @xmath428
. choose holomorphic coordinate charts identifying neighborhoods @xmath429 of each @xmath430 with the standard unit disk @xmath431 such that @xmath432 is mapped to @xmath0 , and use the transformation @xmath433 to define holomorphic cylindrical coordinates @xmath434 on @xmath435 .
since the degree of the cover is finite , each lift to @xmath203 of a loop of the form @xmath436 for @xmath437 closes up after some finite number @xmath59 of iterations , so that the corresponding connected component of @xmath438 can be identified with @xmath439 and @xmath26 on this component takes the form @xmath440 rescaling the @xmath441-coordinate on @xmath438 to turn this map into @xmath442 , @xmath26 then pulls @xmath99 back to the standard complex structure on @xmath443 , so that @xmath203 is now presented as a compact riemann surface with cylindrical ends and can thus be identified with a punctured riemann surface @xmath427 .
assume @xmath178 is a @xmath44-fold holomorphic branched cover of closed connected riemann surfaces with branching data @xmath295 as defined in
[ sec : thebigidea ] , having @xmath85 distinct critical values .
recall from definition [ defn : presentation ] the notion of a regular presentation @xmath239 for @xmath26 , where in particular @xmath113 is a finite set containing the critical values of @xmath26 , giving rise to the punctured surfaces @xmath444 where @xmath445 .
[ lemma : regpres ] there exists a natural bijection between the set of isomorphism classes of regular presentations of @xmath26 and the set of pairs @xmath446 where @xmath113 is a finite subset containing the critical values of @xmath26 and @xmath236 is a finite - index normal subgroup @xmath447 that is contained in @xmath237 .
this bijection matches any minimal regular presentation to the smallest possible choice of @xmath238 and largest possible choice of @xmath236 , i.e. the normal core of @xmath237 .
moreover , if @xmath26 is regular and @xmath239 is a minimal regular presentation , then there exists a diffeomorphism @xmath448 such that @xmath449 . given a finite set @xmath113 containing the critical values of @xmath26 , pick a base point @xmath450 and let @xmath451 denote the universal cover , with @xmath452 defined as a space of homotopy classes of paths beginning at @xmath453 , so that @xmath454 acts naturally on @xmath452 as the group of deck transformations for @xmath455 .
lifting loops based at @xmath453 to paths in @xmath203 then defines a homomorphism @xmath456 so that the covering map @xmath457 can be identified with @xmath458 \mapsto \tilde{\pi}(z),\ ] ] where @xmath459 acts on @xmath452 by deck transformations and on @xmath460 via @xmath461 .
we claim that @xmath462 is the normal core of @xmath237 . indeed , selecting a base point @xmath463 to define @xmath464
, we have @xmath465 which obviously contains @xmath466 . changing the base point @xmath467 changes the subgroup @xmath237 by conjugation with arbitrary elements of @xmath468 , and the normal core is the intersection of all these conjugates , which we can now recognize as the intersection of all the stabilizers of the permutation action on @xmath460 , and that is @xmath466 .
suppose @xmath447 is a finite - index normal subgroup contained in @xmath237 , and therefore also in @xmath466 .
then @xmath461 descends to the finite group @xmath235 , giving a homomorphism @xmath469 which is injective if and only if @xmath470 .
we can then define a regular presentation @xmath471 of @xmath26 with @xmath472 as the natural quotient projection @xmath473 and @xmath474 defined via the quotient projection @xmath475 . observe that if we choose @xmath476 and @xmath26 is regular , then @xmath477 is normal and is therefore identical to @xmath236 , so the natural identification of @xmath203 with @xmath478 gives an isomorphism between the covering maps @xmath26 and @xmath472 .
finally , suppose @xmath239 is a regular presentation of @xmath26 , and define the subgroup @xmath479 , which is normal since @xmath260 is regular and has finite index since @xmath480 is finite .
we claim @xmath481 : indeed , any @xmath482 is represented by a loop @xmath218 based at @xmath453 that lifts to a loop @xmath483 in @xmath240 and thus has @xmath44 lifts to @xmath484 in the form @xmath485 for @xmath263 .
we can therefore use @xmath236 to define the regular presentation from the previous paragraph , with @xmath486 acting on @xmath460 via @xmath461 , and we claim that this is isomorphic to @xmath239 . indeed , choosing a base point @xmath487 , the identification @xmath245 provides a bijection @xmath488 \text { for $ w ' \in \varphi^{-1}(w)$},\ ] ] and combining this with the natural identification of @xmath240 with @xmath489 gives an isomorphism of regular presentations . [ lemma : branchpoints ] suppose @xmath239 is a minimal regular presentation of @xmath178 , and let @xmath269 denote the branched cover of closed riemann surfaces provided by lemma [ lemma : closedup ] such that @xmath262 .
then for each @xmath490 and @xmath491 , the branching order of @xmath472 at @xmath492 is the least common multiple of the branching orders of @xmath26 at all @xmath493 .
in particular , @xmath472 and @xmath26 have the same sets of critical values . if @xmath170 is the branching order of @xmath472 at @xmath492 , we can find punctured neighborhoods @xmath494 of @xmath453 and @xmath495 of @xmath492 and identify both with the half - cylinder @xmath439 with coordinates @xmath496 such that @xmath497 .
let @xmath498 denote the group of automorphisms of @xmath472 that fix @xmath492 ; this is necessarily a cyclic group of order @xmath59 , with a generator @xmath499 that acts on @xmath500 as the rotation @xmath501 . then
since @xmath472 is regular , we can restrict the identification @xmath502 to @xmath503 and obtain an identification @xmath504 the connected components of @xmath505 are then in bijective correspondence to the orbits of the @xmath506-action on @xmath243 defined by @xmath244 , with the branching order @xmath507 of each corresponding point @xmath493 given by the number of points in its respective orbit in @xmath243 . by the orbit - stabilizer theorem , all of these numbers
@xmath508 must divide @xmath509 . if @xmath510 is their least common multiple ,
we conclude that @xmath511 acts trivially on @xmath243 , which means @xmath512 is the identity since @xmath244 is injective for the minimal regular presentation , hence @xmath513
. it will be important to understand how the various objects constructed out of a regular presentation vary smoothly under changes in @xmath26 and @xmath99 .
to this end , we shall fix the following data for the remainder of [ sec : prep ] : * @xmath178 is a holomorphic branched cover of degree @xmath6 with branching data @xmath295 ; * @xmath239 is a regular presentation of @xmath26 ; * @xmath317 is a connected smooth banach manifold ; * @xmath514 is an open subset with compact closure ; * @xmath515 is a smooth family of complex structures on @xmath192 that match @xmath99 outside of @xmath516 ; * @xmath517 is a smooth family of diffeomorphisms @xmath518 which restrict to the identity on @xmath516 and are @xmath99-holomorphic near @xmath238 . we shall abbreviate the family of closed riemann surfaces determined by @xmath519 as @xmath520 and denote by @xmath521 the holomorphic branched cover of closed riemann surfaces provided by lemma [ lemma : closedup ] such that @xmath522 .
these choices produce a family of punctured riemann surfaces @xmath523 and we define @xmath524 where we observe that @xmath525 is always well defined and matches @xmath526 near @xmath527 since @xmath528 is holomorphic near @xmath238 .
this makes @xmath529 a smooth family of holomorphic branched covers , where @xmath530 and they restrict to holomorphic covering maps of punctured surfaces @xmath531 , where @xmath532 [ ex : teichmueller ] suppose @xmath238 is the set of critical values of @xmath26 , @xmath533 , @xmath317 is the @xmath316-dimensional open ball @xmath534 , @xmath535 for all @xmath402 , and @xmath518 is chosen to be any smooth family of diffeomorphisms supported near @xmath238 that are holomorphic in a smaller neighborhood of @xmath238 and such that @xmath536 and @xmath537 is an embedding onto an open subset , where @xmath538
. then the branched covers @xmath539 parametrize a neighborhood of @xmath26 in @xmath327 .
[ ex : teichmueller2 ] if @xmath540 represents a simple element of the moduli space @xmath541 defined in appendix [ sec : critical ] and @xmath542 is generic , then one can enhance the previous example as follows to parametrize a neighborhood of @xmath543 in the space @xmath544 .
a neighborhood of @xmath545 in @xmath541 can be identified with a smooth submanifold @xmath546 of @xmath547 , where @xmath548 is the nonlinear cauchy - riemann operator defined on the product of @xmath549 with a teichmller slice @xmath550 through @xmath551 , cf .
appendix [ sec : critical ] . here
@xmath550 is a finite - dimensional smooth family of complex structures on @xmath192 , which can all be arranged to match @xmath551 near @xmath238 .
a neighborhood in @xmath544 is now parametrized by @xmath552 namely via the curves @xmath553 for each @xmath554 , and we associate to these parameters the families @xmath535 and @xmath555 . [ ex : teichmueller3 ] enhancing the previous example one step further , suppose @xmath556 is an infinite - dimensional banach manifold consisting of smooth almost complex structures and we consider a neighborhood of @xmath557 in the universal moduli space @xmath558 such a neighborhood can be identified with a finite - codimensional submanifold @xmath546 in the infinite - dimensional banach manifold @xmath559 , where @xmath560 .
defining @xmath561 and the families @xmath562 and @xmath563 as in example [ ex : teichmueller2 ] , the parameter space @xmath317 is now infinite dimensional .
observe that the branched covers in the family @xmath407 all have essentially the same topological properties , e.g. their branch points and automorphism groups are identical .
it is therefore trivial to extend @xmath239 to a smooth family of regular presentations @xmath564 for @xmath407 , where @xmath565 . by the same reasoning as above
, we can define on @xmath566 a smooth family of complex structures @xmath567 such that @xmath568 becomes a smooth family of holomorphic branched covers , restricting to a smooth family of holomorphic covering maps @xmath569 , defined on the family of punctured riemann surfaces @xmath570 [ ex : regular ] if @xmath26 is regular with @xmath571 , then it admits a canonical minimal regular presentation @xmath239 where @xmath572 , @xmath573 , @xmath574 , and the action @xmath575 of @xmath202 on itself is defined by left multiplication @xmath576 here the identification @xmath577 sends @xmath578 to @xmath579 $ ] , where @xmath580 is the identity element .
the action of @xmath202 on @xmath581 by deck transformations can now be presented as the action via right multiplication @xmath582 ) \mapsto [ ( z , h g^{-1})].\ ] ] notice that any regular presentation in which @xmath244 acts on @xmath243 both transitively and without fixed points is isomorphic to one of this form , since for any @xmath263 , the map @xmath583 defines a bijection that transforms the action by left multiplication into @xmath204 . [ ex : factorization ] the following construction underlies lemma [ lemma : nonfaithful ] : any proper normal subgroup @xmath584 gives rise to a factorization of @xmath178 in the following way .
let @xmath585 denote the set of orbits for the action @xmath586 .
then @xmath587 is a finite group and @xmath204 descends to a homomorphism @xmath588 which acts transitively on @xmath589 .
the regular cover @xmath590 now factors through the obvious projections @xmath591 and @xmath592 is a regular holomorphic cover with automorphism group @xmath587 .
we can thus define @xmath593 \mapsto \pi_h(z),\ ] ] as well as a factorization of @xmath271 by covering maps @xmath594 where the first map is also defined via the obvious quotient projections .
lemma [ lemma : closedup ] implies that @xmath595 and @xmath596 each arise by puncturing closed connected riemann surfaces @xmath597 and @xmath598 respectively , and in particular we obtain a factorization of @xmath26 via holomorphic branched covers @xmath599 with @xmath600 equal to the number of distinct orbits of the @xmath236-action on @xmath243 , hence @xmath601 holds whenever the action of @xmath236 on @xmath243 is nontrivial .
note that @xmath602 inherits from this construction a regular presentation @xmath603 , though it need not be minimal and @xmath238 may contain points that are not critical values of @xmath602 , even if @xmath239 is minimal .
this is the main reason why non - minimal regular presentations have been included in the discussion .
fix a complex vector bundle @xmath604 of rank @xmath605 , and define the rank @xmath328 bundle of complex - antilinear maps @xmath606 recall that a first - order real - linear partial differential operator @xmath607 is then called a cauchy - riemann type operator on @xmath608 if it satisfies the leibniz rule @xmath609 for all @xmath610 and @xmath611 , where @xmath612 .
the space @xmath613 of all such operators is an affine space modelled on the space of smooth real - linear bundle maps @xmath614 .
the pullback of @xmath615 via @xmath178 defines a cauchy - riemann operator @xmath616 where we define two bundles over @xmath617 by @xmath618 and characterize @xmath619 via the relation @xmath620 if @xmath177 is a @xmath4-holomorphic curve with generalized normal bundle @xmath621 , its normal cauchy - riemann operator @xmath191 belongs to @xmath622 , and if @xmath623 , then @xmath212 and @xmath624 .
fixing hermitian bundle metrics @xmath625 and @xmath626 on @xmath608 and @xmath627 respectively , we can integrate real parts of bundle metrics to define real - valued @xmath628-pairings @xmath629 on @xmath630 and @xmath631 , which determines a formal adjoint operator @xmath632 via the relation @xmath633 for all smooth sections @xmath634 and @xmath610 with compact support .
is compact , but the same definition is also valid on punctured domains . ]
viewing @xmath635 as a fredholm operator on sobolev spaces @xmath636 for some @xmath170 and @xmath171 , we can then identify @xmath637 with @xmath638 , which is the @xmath628-orthogonal complement of @xmath639 and is a finite - dimensional space of smooth sections by elliptic regularity . using the riemann - roch formula @xmath640 and computing the algebraic count of branch points @xmath77 from the riemann - hurwitz formula , the ( real ) fredholm indices of @xmath635 and @xmath619 are related by @xmath641 for technical reasons that will emerge in
[ sec : unique ] , we will sometimes want to explicitly exclude the possibility of @xmath635 being complex linear .
note that any @xmath615 has a complex - linear part which is also a cauchy - riemann operator @xmath642 hence its complex - antilinear part is a tensorial bundle map @xmath643 [ defn : notcomplex ] denote by @xmath644 the subset consisting of operators @xmath635 such that the complex - antilinear part @xmath645 is invertible on some fiber .
the space @xmath646 is open and dense in @xmath647 in the @xmath648-topology , as one can always perturb the antilinear part to make it invertible on some open set .
moreover , whenever @xmath649 , we automatically have @xmath650 . recall that a real subspace @xmath651 of a complex vector space @xmath652 is called totally real if @xmath653 . , but we are not assuming this , e.g. we can allow @xmath654 while @xmath655 . ]
[ lemma : notcomplex ] if @xmath649 , then @xmath656 and @xmath657 are totally real subspaces of @xmath630 and @xmath631 respectively . if @xmath658 and @xmath659 , then @xmath660 and thus @xmath661 , but the latter is impossible if @xmath662 is nontrivial and @xmath645 is invertible on any open subset , since unique continuation implies that @xmath662 can not vanish identically on that subset .
the corresponding statement about @xmath663 follows since the antilinear part of @xmath664 is the transpose of the antilinear part of @xmath635 and is thus invertible on any fiber where the latter is invertible . in order to exploit the topological constructions in the previous section , we will need to work with cauchy - riemann type operators on punctured surfaces instead of closed surfaces .
we shall now show that this can be done without loss of generality by choosing suitable weighted sobolev spaces .
assume @xmath665 is a smooth family of rank @xmath328 complex vector bundles with complex structures @xmath666 , equipped with a smooth family of cauchy - riemann operators @xmath667 .
denote the restrictions of the bundles @xmath668 and @xmath669 to the punctured surfaces @xmath670 by @xmath671 restricting @xmath672 to @xmath670 then defines a family of cauchy - riemann type operators @xmath673 in order to understand the functional - analytic properties of @xmath674 , we must examine its asymptotic behavior fairly carefully .
fix local holomorphic coordinate charts to identify a neighborhood of each @xmath490 in @xmath192 with the closed unit disk @xmath431 , with @xmath453 corresponding to @xmath675 , and use @xmath528 to produce from these a smooth family of holomorphic charts on neighborhoods of @xmath676 for @xmath677 . in these coordinates , use the biholomorphic map @xmath678 to define cylindrical ends of @xmath670 with holomorphic coordinates @xmath434 .
choose also a smooth family of trivializations of @xmath668 near @xmath679 and denote the resulting trivialization of @xmath680 over the cylindrical ends by @xmath681 .
the relative first chern number over a surface @xmath192 with a trivialization @xmath681 specified outside of some open subset in @xmath192 with compact closure , the relevative first chern number @xmath682 is defined by algebraically counting the zeroes of a generic section that is constant with respect to @xmath681 wherever the latter is defined .
this definition extends uniquely to higher rank bundles via the relation @xmath683 .
] of @xmath680 is then given by @xmath684 for any tuple of real numbers @xmath685 we can use the chosen coordinates and trivializations over the cylindrical ends of @xmath670 to define the sobolev space with exponential weights @xmath686 we will also write @xmath687 note that sections @xmath688 have exponential decay at any end where @xmath689 , but one can also take @xmath690 , in which case @xmath662 may be unbounded with exponential _ growth _ near @xmath453 . in order to emphasize when we are using negative exponential weights
, we associate to @xmath691 the inverse set of weights @xmath692 the asymptotic coordinates and trivializations also naturally give rise to asymptotic trivializations of @xmath693 , so we can similarly define the banach space @xmath694 , which is a completion of some subset of @xmath695 determined by the asymptotic conditions .
choose a smooth @xmath402-parametrized family of hermitian bundle metrics and connections on @xmath668 which match the trivial metric and connection in our chosen family of trivializations near @xmath679 .
any cauchy - riemann type operator on @xmath668 can then be written as @xmath696 for some @xmath697 , where @xmath698 . in the chosen
coordinates and trivialization near a point @xmath699 , the @xmath158-form @xmath700 can be written as @xmath701 for some smooth function @xmath702 .
the restriction of @xmath700 to an @xmath703-valued @xmath158-form @xmath704 can then be written on the corresponding cylindrical end as @xmath705 where @xmath706 and given a section @xmath707 expressed as a function @xmath708 with respect to the trivialization on the same end , @xmath709 on this end takes the form @xmath710 observe that @xmath711 with all derivatives as @xmath712 .
this expression shows that @xmath674 extends to a bounded linear operator @xmath713 for any choices of @xmath170 , @xmath171 and exponential weights @xmath714 .
operators of this type are standard in floer - type theories and especially in symplectic field theory . appealing to the fredholm theory on punctured surfaces developed in @xcite , the asymptotic decay of @xmath715
means that @xmath716 behaves like a linearized cauchy - riemann operator for a punctured holomorphic curve with trivial asymptotic operators ( corresponding to degenerate periodic orbits ) , and is thus _ not _ fredholm . but
it becomes fredholm when we introduce suitable weights : conjugating @xmath717 with a map of the form @xmath718 for a suitable function @xmath719 ( cf .
@xcite*6 or @xcite*2.1 ) produces a commutative diagram @xmath720 where @xmath721 is another cauchy - riemann type operator whose asymptotic operators are offset by constants depending on the weights @xmath722 , and thus is fredholm for suitable choices . in particular ,
the computation in and below will show that imposing the exponential growth condition @xmath723 on each cylindrical end for sufficiently small @xmath724 adjusts the asymptotic operators of @xmath725 so that each acquires an effective conley - zehnder index of @xmath328 relative to the trivialization @xmath681 .
we need to be a bit cautious with the weights when discussing elliptic regularity and formal adjoints : as a rule , the sobolev constants @xmath170 and @xmath171 can be changed freely , but the weights can not .
the following are immediate consequences of after applying standard regularity arguments to @xmath725 , plus the fact cauchy - riemann operators with nondegenerate asymptotics automatically impose exponential decay conditions on their kernels ( cf .
@xcite*prop .
3.1.26 ) : [ lemma : regularity ] suppose @xmath170 , @xmath726 , and @xmath727 is any choice of exponential weights .
if @xmath728 is a weak solution to @xmath729 for @xmath730 , then @xmath688 .
[ lemma : regularity2 ] suppose @xmath726 and the weights @xmath722 are chosen such that @xmath731 is fredholm . if @xmath732 is a weak solution to @xmath733 , then @xmath734 for all @xmath170 and @xmath735 . to discuss the formal adjoint on punctured domains ,
one should define real @xmath628-products for @xmath736 and @xmath737 in terms of a family of hermitian bundle metrics on @xmath668 and riemannian metrics on @xmath670 that are compatible with the conformal structure and standard on the cylindrical ends .
the key technical point is then the following : there are well - defined @xmath628-pairings @xmath738 whenever @xmath739 , and using the density of @xmath740 , the usual relation @xmath741 for smooth sections with compact support remains valid whenever @xmath742 and @xmath743 for @xmath739 . using ,
one finds @xmath744 , from which one can check that @xmath745 satisfies the fredholm property and lemmas [ lemma : regularity ] and [ lemma : regularity2 ] under the same conditions on @xmath722 as @xmath746 .
the next result appears standard at first glance , but the reader should be cautioned that it depends on inclusions @xmath747 which hold only when all the weights are nonnegative , so e.g. one does not obtain any similar result with the roles of @xmath674 and @xmath748 reversed .
[ prop : formaladjoint ] assume @xmath170 , @xmath726 , and @xmath749 is a set of nonnegative exponential weights such that @xmath750 is fredholm . defining its formal adjoint as a bounded linear map @xmath751 and using the obvious inclusions @xmath752 , we have @xmath753 in particular , @xmath754 is isomorphic to the space of all sections in @xmath755 for @xmath739 that are @xmath628-orthogonal to @xmath756 under the pairing . if @xmath757 , then @xmath758 for some @xmath759 , while @xmath760 also belongs to @xmath761 for @xmath739 by lemma [ lemma : regularity2 ] .
thus @xmath760 has a well - defined @xmath628-pairing with itself and gives @xmath762 to show that @xmath763 is @xmath764 , note first that it is a closed subspace since @xmath674 is fredholm .
then in the case @xmath765 , the contrary would mean there exists a nontrivial @xmath766 for @xmath739 such that @xmath767 for all @xmath742 and @xmath768 for all @xmath769 .
the first condition means @xmath770 by lemma [ lemma : regularity2 ] and thus contradicts the second unless @xmath771 . to extend this result to all @xmath170 ,
note that if @xmath772 then the @xmath765 case gives @xmath742 and @xmath773 such that @xmath774 .
then lemma [ lemma : regularity2 ] implies @xmath775 , implying that @xmath709 is also in @xmath764 , so lemma [ lemma : regularity ] implies @xmath776 and we are done .
this discussion extends easily to the pulled back operators @xmath777 on bundles over @xmath778 and @xmath779 respectively .
observe that since @xmath780 has no branch points , @xmath781 gives a bundle isomorphism @xmath782 and we can thus identify @xmath783 so that @xmath784 can be viewed as a map @xmath785 .
we can now define fixed holomorphic cylindrical coordinate systems @xmath434 on punctured neighborhoods of each point @xmath786 such that @xmath407 takes the form @xmath787 where @xmath788 is the branching order of @xmath26 at @xmath492 . pulling back the trivializations @xmath681 on @xmath668 near @xmath679 to define corresponding trivializations of @xmath789 near @xmath527
, we obtain asymptotic trivializations of @xmath790 and @xmath791 on the cylindrical ends and can thus define weighted sobolev norms for sections of these bundles , producing a bounded linear operator @xmath792 for all choices of @xmath170 , @xmath171 and exponential weights @xmath793 . if @xmath794 is a choice of weights for @xmath674 , there is an induced set of weights for @xmath795 defined by @xmath796 where @xmath797 again denotes the branching order of @xmath26 at @xmath492 .
[ prop : punctured ] suppose @xmath170 , @xmath171 , and the exponential weights @xmath794 are chosen to satisfy @xmath798 for every @xmath490 .
then for any @xmath667 , the operators @xmath799 are fredholm and satisfy @xmath800 moreover , the maps @xmath801 and @xmath802 defined by restricting smooth sections to the corresponding punctured domains define isomorphisms @xmath803 we will prove the correspondence between @xmath672 and @xmath674 , as the result for the pulled back operators follows by the same argument simply replacing the bundles @xmath804 and @xmath805 with @xmath806 and @xmath807 respectively .
the fredholm property for @xmath674 and the index calculation follow from the usual index formula for cauchy - riemann operators on riemann surfaces with cylindrical ends , proved in @xcite , supplemented by the transformation to handle the exponential weights ( cf .
@xcite*6 ) .
in particular , the condition @xmath808 for each @xmath699 guarantees that @xmath674 is conjugate ( cf . and below ) to a cauchy - riemann type operator @xmath809 with nondegenerate asymptotic operators at every puncture whose conley - zehnder indices with respect to the trivialization @xmath681 are @xmath810 . in light of ,
the index formula from @xcite thus gives @xmath811 note that doing the same computation for the pulled back operators requires the stronger condition @xmath812 in order to ensure that all of the pulled back weights in the set @xmath813 lie in the interval @xmath814 .
to understand the kernels , observe that since any @xmath815 is smooth , its restriction to @xmath670 belongs to @xmath816 and is thus in @xmath817 . would not belong to @xmath816 in general if @xmath662 were an arbitrary ( not necessarily smooth ) section of class @xmath818 on @xmath668 , nor if any of the exponential weights were nonnegative
the latter in particular permits sections in @xmath816 that do not decay to zero at infinity , which is crucial since arbitrary smooth sections @xmath815 may indeed by nonzero at points in @xmath679 . ]
conversely , we need to show that any section @xmath776 annihilated by @xmath674 can be extended over the punctures to a section in @xmath819 , which is then automatically annihilated by @xmath672 .
this will follow from the asymptotic elliptic theory of the equation @xmath733 .
indeed , recall from that on the cylindrical end near any puncture @xmath699 , the function @xmath708 representing @xmath820 in some trivialization satisfies @xmath821 and @xmath822 where @xmath823 .
then @xmath824 satisfies the cauchy - riemann type equation @xmath825 f = 0.\ ] ] since @xmath711 as @xmath712 , this equation is asymptotic to the equation @xmath826 for the asymptotic operator @xmath827 which can be regarded as a densely defined unbounded self - adjoint operator on @xmath828 .
the function @xmath702 is smooth by assumption , and then implies that the derivatives @xmath829 of @xmath830 for arbitrary multi - indices @xmath760 satisfy exponential decay conditions @xmath831 for suitable constants @xmath832 .
applying @xcite*theorem a.1 , @xmath833 therefore satisfies @xmath834,\ ] ] where @xmath835 is a nontrivial eigenfunction of @xmath836 with eigenvalue @xmath837 , and the remainder @xmath838 decays to zero with all its derivatives uniformly in @xmath839 as @xmath712 .
the spectrum of @xmath836 is @xmath840 , hence the assumption @xmath841 implies @xmath842 , and we conclude that @xmath843\ ] ] is bounded on the cylindrical end ; in fact , one can use this to show that the smooth function @xmath844 defined via the transformation @xmath845 has finite @xmath846-norm on @xmath847 .
moreover , @xmath848 has a continuous extension to @xmath849 : indeed , the extension is obviously @xmath850 if @xmath851 , while in the case @xmath852 , the eigenfunction @xmath853 is necessarily constant , so that @xmath854 converges to this constant value as @xmath712 .
all these conditions together imply that the continuous extension of @xmath662 over the punctures is of class @xmath818 , e.g. the case @xmath765 is a standard exercise using the definition of weak derivatives ( cf .
@xcite*exercise 2.118 ) , and the general case follows from this by elliptic regularity . since sections in @xmath816 and
its pulled back counterpart need not be bounded when the weights @xmath855 are negative , the punctured operators in proposition [ prop : punctured ] can not be interpreted in any reasonable way as linearizations of nonlinear cauchy - riemann operators , e.g. @xmath816 in this case is not a subspace of a tangent space in any reasonable banach manifold . for our purposes ,
the exponential growth condition is merely a technical convenience so that we can consider operators with the right index and the right kernel and cokernel while dealing with honest covering maps instead of branched covers .
the geometrically meaningful operators are still @xmath672 and @xmath856 , on unpunctured domains .
suppose @xmath668 , @xmath857 and @xmath672 are independent of @xmath402 but @xmath407 moves in @xmath327 as @xmath402 varies , e.g. this is the relevant situation for the proof of super - rigidity .
there is then a subtle but important difference between what proposition [ prop : punctured ] says about @xmath674 and what it says about @xmath795 .
the former is a family of operators whose relationship to each other for different values of @xmath402 is not obvious from the definitions , but the proposition implies that they are all in some sense equivalent to a single operator @xmath635 on the closed domain , so they all have isomorphic kernels .
no such thing can be assumed for the pulled back operators : while @xmath795 must have the same index for all @xmath402 , there is nothing in this setup to stop the dimension of its kernel from varying wildly with @xmath402 . in preparation for the twisted bundle construction in the next section
, we now collect some general facts from representation theory . given a finite set @xmath243 with @xmath858 elements and a finite group with a homomorphism @xmath859 defining a transitive group action on @xmath243 , we denote by @xmath273 the real vector space spanned by basis vectors @xmath860 , with an inner product such that this basis is orthonormal .
we shall use the boldface symbol @xmath366 to denote the corresponding real @xmath44-dimensional representation of @xmath202 , @xmath861 we will be interested in the decomposition of @xmath273 into irreducible @xmath202-invariant summands .
this is easiest to understand in terms of its complexification @xmath862 defined by viewing @xmath860 as a complex basis of @xmath863 . in general
, we say that a complex representation @xmath864 is the complexification of a real representation @xmath278 if @xmath865 is isomorphic to @xmath866 such that @xmath202 acts on the latter by the complex - linear extension of its action on @xmath651 .
recall from @xcite*13.2 that irreducible complex representations @xmath867 come in three mutually exclusive types : * real type : @xmath865 admits a complex - antilinear @xmath202-invariant involution . then @xmath868 is the complexification of a real irreducible representation @xmath278 .
it follows that @xmath868 is isomorphic to its dual representation @xmath869 , and all @xmath202-equivariant linear maps @xmath289 are given by scalar multiplication : @xmath870 * complex type : @xmath868 is not isomorphic to its dual representation @xmath871 .
then @xmath872 is the complexification of a real irreducible representation @xmath278 obtained from @xmath867 by setting @xmath873 and using the obvious inclusion @xmath874 .
the space of @xmath202-equivariant real - linear maps on @xmath651 is then real @xmath195-dimensional : @xmath875 * quaternioic type : @xmath868 is not of real type but is nonetheless isomorphic to its dual representation .
then @xmath876 is the complexification of a real irreducible representation @xmath278 obtained from @xmath867 by setting @xmath873 and using the obvious inclusion @xmath874 , and the space of @xmath202-equivariant real - linear maps on @xmath651 is real @xmath1-dimensional , i.e. it is isomorphic to the quaternions : @xmath877 we shall also refer to a real irreducible representation as `` of real / complex / quaternioic type '' according to which of these three constructions it comes from . with this classification in mind , we denote the various complex irreducible representations of @xmath202 by @xmath878 where @xmath879 stands for @xmath880 , @xmath881 or @xmath882 depending on the type , and arrange a complete list of pairwise non - isomorphic irreducible representations in the form @xmath883 this gives rise to a corresponding complete list @xmath884 of pairwise non - isomorphic real irreducible representations @xmath885 where for each @xmath99 , the complexification of @xmath886 is @xmath887 for @xmath888 , @xmath889 for @xmath890 , and @xmath891 for @xmath892 . since @xmath893 itself is a complexification of a real representation , every subspace in the resulting isotypic decomposition of @xmath863 is either identical or orthogonal to its complex conjugate , where the conjugate always carries the dual representation .
thus we can uniquely decompose @xmath863 into pairwise orthogonal @xmath202-invariant complex subspaces @xmath894 where each @xmath895 and @xmath896 is of the form @xmath897 for some real subspace @xmath898 , and each @xmath899 has trivial intersection with @xmath273 .
next , observe that every irreducible @xmath202-invariant subspace in @xmath863 is either identical to its complex conjugate or intersects it trivially : indeed , any other option would produce an intersection which is a nontrivial but smaller @xmath202-invariant subspace .
we can thus further decompose @xmath895 and @xmath899 into irreducible @xmath202-invariant subspaces @xmath900 for some integers @xmath901 , where each @xmath902 summand in @xmath895 can be assumed of the form @xmath903 for some irreducible @xmath202-invariant real subspace @xmath904 . in @xmath896 , the irreducible @xmath202-invariant subspaces can not be complexifications since the corresponding representation is not realizable over @xmath880 , thus these subspaces have trivial intersection with @xmath273 and can instead be arranged in conjugate pairs : @xmath905 for some integers @xmath906 . from this decomposition of @xmath893
we can immediately read off a corresponding decomposition of @xmath366 : we have @xmath907 where the summands are all @xmath202-invariant and pairwise orthogonal , @xmath908 for @xmath909 , and @xmath910 .
these summands then admit further ( non - unique ) decompositions into real irreducible @xmath202-invariant subspaces @xmath911 we now specialize the above discussion to the case @xmath912 in which case @xmath913 is the so - called regular representation of @xmath202 . by a standard theorem in complex representation theory , the complexification @xmath914 then contains every irreducible complex representation @xmath915 as a subrepresentation with multiplicity equal to @xmath916 .
this implies a similar fact about @xmath366 that we will make use of in [ sec : sardsmale ] for proving theorem [ thm : submanifolds0 ] : [ lemma : regular ] the real regular representation @xmath913 contains every irreducible real representation @xmath917 of @xmath202 as a subrepresentation with positive multiplicity .
next , recall that the action of @xmath202 on itself by right multiplication @xmath918 commutes with @xmath204 and thus defines a second permutation representation @xmath919 which commutes with @xmath366 , giving rise to a representation @xmath920 by another standard theorem of complex representation theory , the summands in the isotypic decomposition of @xmath921 are then invariant under the complexification of the @xmath922-action , and they define irreducible complex representations of @xmath923
. in particular , @xmath924 therefore preserves each isotypic component for @xmath366 but does not preserve any further decomposition of that component into irreducible @xmath202-invariant subspaces . for future use , we note one additional fact from complex representation theory : the action of @xmath923 on an isotypic component in @xmath921 corresponding to a given irreducible representation @xmath867 is isomorphic to @xmath925 , with @xmath923 acting by @xmath926 cf .
@xcite*6.2 .
an important special case of the factorization construction in example [ ex : factorization ] arises when @xmath927 is a real irreducible representation that is not faithful .
choosing @xmath236 to be any nontrivial normal subgroup of its kernel @xmath928 @xmath587 then inherits an irreducible representation @xmath929 for example one can take @xmath930 , in which case @xmath931 becomes faithful .
now if @xmath244 is a transitive action on the set @xmath243 of @xmath44 elements , let @xmath932 denote the induced action on the set @xmath585 of @xmath236-orbits , and consider the corresponding permutation representations @xmath933 [ lemma : characters ] under the assumptions described above , the multiplicity of @xmath278 as a subrepresentation of @xmath934 matches the multiplicity of @xmath935 as a subrepresentation of @xmath936 .
observe that in terms of the real / complex / quaternionic distinction described in
[ sec : realcomplex ] , @xmath283 and @xmath931 are necessarily of the same type : indeed , the spaces of linear maps on @xmath651 that are @xmath202-equivariant or @xmath937-equivariant are the same since @xmath236 acts trivially on @xmath651 .
the multiplicities of both are therefore determined in the same way by the multiplicities of the corresponding _ complex _ irreducible representations in the complexifications of @xmath366 and @xmath938 respectively , thus it will suffice to prove a similar statement about complex representations .
namely , assume @xmath867 is complex irreducible , @xmath939 is a normal subgroup and @xmath940 is the resulting irreducible representation of @xmath587 . by orthonormality of characters
, it will suffice to prove @xmath941 where the inner product of characters @xmath942 is given in general by @xmath943 for each @xmath263 , let @xmath264 denote the stabilizer subgroup for @xmath265 under the @xmath202-action on @xmath243 via @xmath366 .
since the action is transitive , the orbit - stabilizer theorem implies @xmath944 .
the trace of a permutation matrix is the number of elements that it fixes , in other words the number of stabilizer subgroups that it belongs to , hence for each @xmath945 , @xmath946 this implies @xmath947 this can be simplified since @xmath202 acts transitively on @xmath243 , so the subgroups @xmath948 for distinct @xmath263 are all conjugate . by the conjugation - invariance of characters , this implies that all @xmath44 of the sums over @xmath948 in are identical , so plugging in @xmath944 , we have @xmath949 where @xmath263 in this expression can be chosen arbitrarily . to write down a similar expression for @xmath950 , define for each @xmath263 @xmath951 which is a subgroup of both @xmath236 and @xmath948 and is normal in the latter .
there is then a natural inclusion of @xmath952 as a subgroup of @xmath587 , and it is the stabilizer subgroup of @xmath953 \in i / h$ ] for the permutation action of @xmath587 on @xmath589 .
the same computation thus gives @xmath954 \in g_i / h_i } \chi_{{\boldsymbol{\lambda}}_h}([g ] ) = \frac{|h_i|}{|g_i| } \sum_{[g ] \in g_i / h_i } \chi_{{\boldsymbol{\lambda}}_h}([g]).\ ] ] finally , observe that @xmath955)$ ] for each @xmath945 since both are traces of the same matrix acting on @xmath865 , so one can replace the last expression with a sum over @xmath956 , giving @xmath957 we can now make precise the splitting of pulled back cauchy - riemann type operators that was sketched in [ sec : thebigidea ] .
we associate to any real representation @xmath278 the family of real vector bundles @xmath958 defined by @xmath959 where @xmath202 acts on @xmath651 via @xmath283 and on @xmath960 by deck transformations , so that @xmath961 identifies the latter with @xmath962 .
this gives rise to complex vector bundles @xmath963 of rank @xmath964 , defined by @xmath965 each of the bundles @xmath966 has a canonical flat structure , i.e. it comes with a well - defined notion of locally constant sections , thus @xmath967 determines a family of cauchy - riemann type operators @xmath968 such that @xmath969 whenever @xmath165 is a locally constant section of @xmath966 . since @xmath970 , it is fredholm in suitable banach space settings , in particular as a bounded linear operator @xmath971 for any @xmath170 , @xmath171 , and negative exponential weights @xmath972 with all @xmath689 sufficiently small .
we will formulate a precise version of this statement and compute the index in [ sec : index ] .
observe that aside from its obvious dependence on @xmath672 , @xmath973 depends on our choice of regular presentation for @xmath26 and on the representation @xmath283 , but both of them only up to isomorphism .
the most important special case of the above construction is @xmath974 , where @xmath975 is the permutation representation associated to our regular presentation of @xmath26 .
we define @xmath976 as above and can identify it canonically with @xmath977 so that sections of @xmath978 can be written as @xmath202-equivariant sections of @xmath979 , hence @xmath980 for @xmath981 . here
@xmath202-equivariance means that for all @xmath982 and @xmath945 , @xmath983 hence @xmath984 writing @xmath985 , this relation gives rise to a bijective correspondence @xmath986 ) = \eta^i(z ) \end{split}\ ] ] and thus natural isomorphisms @xmath987 for every @xmath988 and @xmath171 , where the exponential weights @xmath989 and @xmath990 can all be assumed arbitrarily close to zero .
observe that @xmath991 also has a well - defined real bundle metric since @xmath366 acts on @xmath273 by orthogonal transformations , so endowing @xmath668 with a hermitian bundle metric induces a hermitian bundle metric on @xmath992 such that the correspondence also preserves @xmath628-products .
after writing down a similar correspondence for the bundles @xmath993 and @xmath791 , we obtain an identification between the cauchy - riemann operators @xmath994 and @xmath995 : @xmath996 if @xmath997 is any @xmath202-invariant subspace and @xmath278 denotes the resulting subrepresentation , then we obtain corresponding subbundles @xmath998 such that @xmath995 takes sections of @xmath999 to sections of @xmath1000 , acting as the operator @xmath973 . under the correspondence , one can understand this as identifying @xmath1001 and @xmath1002 with closed subspaces @xmath1003 with a similar definition for closed subspaces of the relevant weighted sobolev spaces , such that @xmath795 restricts to a bounded linear operator @xmath1004 which is conjugate to @xmath1005 and will thus be fredholm with any negative exponential weights that are close enough to @xmath0 .
now if @xmath1006 is a decomposition of @xmath1007 into subrepresentations @xmath1008 for @xmath1009 , we obtain a direct sum decomposition @xmath1010 which is equivalent via to a decomposition of @xmath795 over a splitting of banach spaces @xmath1011 and the corresponding decomposition of @xmath1012 .
note that if the subspaces @xmath1013 are pairwise orthogonal , then the corresponding spaces of sections of @xmath790 and @xmath791 are @xmath628-orthogonal as a consequence .
it is useful to observe that whenever two of the representations @xmath1014 and @xmath1008 are isomorphic , the @xmath202-equivariant isomorphism @xmath1015 induces bundle isomorphisms @xmath1016 and @xmath1017 that identify @xmath1018 with @xmath1019 , so these two operators have isomorphic kernels and cokernels .
this implies : [ lemma : kerneldecomp ] suppose @xmath1008 for @xmath1009 is a set of pairwise non - isomorphic real representations of @xmath202 , and @xmath278 is another representation such that @xmath1020 for some integers @xmath1021
. then there exist isomorphisms @xmath1022 in particular , if @xmath283 is the permutation representation @xmath934 , this gives isomorphisms @xmath1023 here is a quick proof of lemma [ lemma : nonfaithful ] .
for the present discussion we drop the parameter @xmath402 from the notation since it does not play any important role .
suppose @xmath278 is a representation and @xmath1024 is a nontrivial normal subgroup , giving rise to a representation @xmath1025 and ( following example [ ex : factorization ] ) a factorization of @xmath234 as @xmath1026 by assumption we are using a minimal regular presentation and thus @xmath244 is injective , so @xmath236 acts nontrivially on @xmath243 , implying @xmath1027 .
writing @xmath1028 , the obvious projection map @xmath1029 is then an isomorphism of real vector bundles over @xmath218 and thus gives rise to a canonical identification between the twisted bundles @xmath1030 and @xmath1031 with their cauchy - riemann operators @xmath1032 and @xmath1033 . to prove the lemma , we now just need to observe that lemma [ lemma : characters ] implies @xmath283 is a subrepresentation of @xmath366 if and only if @xmath931 is a subrepresentation of @xmath938 , hence the corresponding twisted operators appear simultaneously as summands in the decompositions of @xmath1034 and @xmath1035 from lemma [ lemma : kerneldecomp ] .
in the situation above , one should interpret @xmath1036 as the set of all sections in @xmath1037 that are pullbacks of sections in @xmath1038 ( interpreted as a subspace of @xmath1039 ) via the branched cover @xmath1040 .
now consider the special case where @xmath366 is the regular representation @xmath1041 , defined via @xmath1042 we saw in example [ ex : regular ] that this means @xmath1043 are all regular covers isomorphic to @xmath1044 , and the action of @xmath202 on @xmath1045 by deck transformations takes the form @xmath1046 : = [ ( z,\rho'_g(h))]\ ] ] where @xmath1047 is the action of @xmath202 on itself by right multiplication , @xmath1048 . the induced @xmath202-action on spaces of sections @xmath662 of @xmath790
is defined by @xmath1049 ) : = \eta(g^{-1 } [ ( z , h ) ] ) = \eta([(z , hg)]).\ ] ] recall now from [ sec : regular ] that the permutation representation @xmath1050 arising from @xmath1051 commutes with @xmath366 and preserves the isotypic components of @xmath366 .
it therefore defines an action on @xmath978 by fiber - preserving bundle isomorphisms , and these isomorphisms preserve each of the subbundles in the splitting @xmath1052 corresponding to the isotypic decomposition of @xmath1007 .
in particular , this @xmath202-action by bundle isomorphisms gives a linear @xmath202-action on each of the subspaces @xmath1053 , and there is a similar action on sections of @xmath993 such that the restriction of @xmath995 to each of these subspaces is @xmath202-equivariant .
its kernel and cokernel thus inherit natural @xmath202-actions . under the correspondence
, this action on sections of @xmath1054 matches the action by deck transformations on @xmath1055 .
[ lemma : ndimension ] suppose @xmath575 is defined by left multiplication , @xmath1056 is a real irreducible representation of @xmath202 , and @xmath1057 denotes the corresponding summand in the isotypic decomposition of the regular representation @xmath1058 .
then every irreducible subrepresentation for the natural @xmath202-action on @xmath1059 or @xmath1060 is isomorphic to @xmath1061 .
suppose first that @xmath1061 is of either real or quaternionic type , in which case the complexification @xmath1062 of @xmath1063 is also an isotypic component for the complexified regular representation @xmath914 .
we shall denote the restriction of @xmath893 to @xmath546 by @xmath1064 and let @xmath1065 denote the underlying complex irreducible representation .
regarding these complex representations as real representations on @xmath546 and @xmath865 respectively gives rise to corresponding twisted bundles and cauchy - riemann operators on them , along with a natural linear inclusion of vector bundles @xmath1066 it will be useful to think of @xmath1067 as a _ complexification _ of @xmath1068 , in the following sense . while @xmath1068 is already a complex vector bundle
, @xmath1069 naturally carries _ two _ complex structures @xmath666 and @xmath265 , which commute with each other : the former acts on @xmath1070 by @xmath1071 and the latter by @xmath1072 , using the fact that @xmath1073 is a complex representation and @xmath1074 is therefore naturally a complex vector bundle . from this perspective , @xmath1075 is the natural @xmath265-complex - linear extension of @xmath1076 to its complexified domain , and the representations defined by the @xmath202-action on @xmath1077 and @xmath1078 will be the complexifications of the real representations it defines on @xmath1079 and @xmath1080 respectively . in the following
we shall use the symbol `` @xmath1081 '' to denote complex tensor products of vector spaces and bundles with @xmath265 ( instead of @xmath666 ) as the complex structure .
recall now that as an isotypic component of the complex regular representation , @xmath546 admits a complex - linear isomorphism to @xmath1082 such that for all @xmath945 , @xmath1083 acts on @xmath1082 as @xmath1084 , while @xmath1085 acts as @xmath1086 .
the isomorphism @xmath1087 thus gives rise to @xmath265-complex bundle isomorphisms @xmath1088 where we are abusing notation to let @xmath1089 denote the trivial bundle over @xmath670 with fiber @xmath1089 , and this identifies @xmath1075 with @xmath1090 .
we therefore have @xmath1091 with @xmath202 acting on both by @xmath1092 , hence all irreducible subrepresentations in these spaces are isomorphic to @xmath1093 , which is isomorphic to @xmath1094 since the latter is not of complex type . viewing these as complexifications of real representations on @xmath1079 and @xmath1080 as explained above
, it follows via the correspondence between real and complex irreducible representations outlined in [ sec : realcomplex ] that all the irreducible real subrepresentations are isomorphic to @xmath1061 .
the main difference if @xmath1061 is of complex type is that @xmath1095 is no longer an isotypic component for @xmath893 , but is instead the direct sum of two isotypic components related to each other by complex conjugation @xmath1096 corresponding to some complex irreducible representation @xmath1065 and its non - isomorphic dual @xmath1097 . writing @xmath1098 and @xmath1099 for the restriction of @xmath893 to these subspaces , we can then think of @xmath1100 as the complexification of @xmath1076 .
a repeat of the argument above using the isomorphisms @xmath1101 and @xmath1102 then gives an @xmath265-complex - linear isomorphism @xmath1103 with @xmath202 acting via @xmath1092 on the first summand and @xmath1104 on the second , and a similar isomorphism for cokernels .
it follows that every irreducible subrepresentation in either @xmath1105 or @xmath1106 is isomorphic to one of @xmath1094 or @xmath1107 , and the desired result for real subrepresentations again follows via the correspondence between real and complex representations in [ sec : realcomplex ] .
combining this with lemma [ lemma : kerneldecomp ] and applying schur s lemma now gives : [ cor : codimension ] in the setting of lemma [ lemma : ndimension ] , let @xmath1108 and @xmath1109 .
then the space of @xmath202-equivariant real - linear maps @xmath1110 has real dimension @xmath1111 we assume throughout this section that @xmath239 is the minimal regular presentation of @xmath234 .
suppose @xmath1112 is a complete list of pairwise non - isomorphic real irreducible representations for @xmath202 .
any @xmath333-tuples of nonnegative integers @xmath334 and @xmath335 now determine subsets of the parameter space @xmath1113 note that @xmath1114 is automatically empty unless @xmath1115 for all @xmath341 , and these indices do not depend on the parameter @xmath402 . assuming this condition holds
, we would now like to present @xmath1114 locally as the zero - set of a smooth map to a finite - dimensional vector space , and to compute its derivative in a special case .
we start by translating the conditions defining @xmath1114 into conditions on the pulled back operators @xmath1116 for a suitable family of _ regular _ covers @xmath1117 with @xmath1118 .
this can be defined by replacing the homomorphism @xmath244 with the action of @xmath202 on itself by left multiplication , i.e. let @xmath1119 so that @xmath1120 becomes a minimal regular presentation for @xmath1121 \mapsto \pi_\tau(z),\ ] ] or rather for the extension of this map to a branched cover of closed surfaces as provided by lemma [ lemma : closedup ] . in keeping with our usual notational convention , @xmath1122 is a fixed smooth surface @xmath1123 with a fixed @xmath202-action by deck transformations but a @xmath402-dependent family of conformal structures @xmath1124 , which are fixed on the cylindrical ends . denote the isotypic decomposition of the regular representation @xmath1125 by @xmath1126 where @xmath1127 for some integers @xmath322 which are strictly positive by lemma [ lemma : regular ] .
then by lemma [ lemma : kerneldecomp ] , @xmath1128 so @xmath1129 implies @xmath1130 [ lemma : localv ] every @xmath1131 has a neighborhood @xmath1132 such that @xmath1133 is the set of all @xmath1134 for which holds .
since all the operators @xmath1018 are fredholm and they depend continuously on @xmath402 , we can assume @xmath1135 for all @xmath341 if @xmath402 is sufficiently close to @xmath1136 .
thus can only be satisfied if none of these inequalities are strict , which means @xmath1129 since every @xmath322 is positive .
recall from
[ sec : crpunctured ] that the weighted sobolev spaces @xmath1137 and @xmath1138 are defined in terms of fixed families of trivializations of @xmath668 near @xmath679 and holomorphic cylindrical coordinates which allow us to compute sobolev norms on the cylindrical ends . given @xmath1131 ,
choose a neighborhood @xmath1139 that is diffeomorphic to a ball and small enough to satisfy lemma [ lemma : localv ] .
by assumption the bundles @xmath668 depend smoothly on @xmath402 , which means there is a well - defined smooth bundle @xmath1140 with @xmath1141 . choosing a suitable connection on the latter , we can use parallel transport along paths of the form @xmath1142 with @xmath1143 radiating outward from @xmath1136 to define a smooth family of complex bundle isomorphisms @xmath1144 which respect these fixed trivializations near @xmath679 and satisfy @xmath1145 .
these give rise to isomorphisms @xmath1146 covering the diffeomorphisms @xmath1147 .
notice that there are also natural real bundle isomorphisms @xmath1148 so that @xmath1149 gives a family of isomorphisms @xmath1150 covering @xmath1151 , and they respect the chosen holomorphic cylindrical coordinates on the ends .
these then induce smooth families of isomorphisms of complex bundles over @xmath1123 , @xmath1152 which again are the identity for @xmath1153 and are also equivariant with respect to the natural @xmath202-action by bundle isomorphisms covering deck transformations of @xmath1123 . acting with
these on sections produces @xmath402-parametrized families of @xmath202-equivariant banach space isomorphisms which we shall also denote by @xmath1154 : @xmath1155 here @xmath1145 .
we can now use these isomorphisms to define for @xmath1156 a smooth family of @xmath202-equivariant fredholm operators with fixed domain and target space , @xmath1157 such that @xmath1158 in order to present the latter as the zero - set of a smooth map , let us abbreviate @xmath1159 so defines a smooth map @xmath1160 where @xmath1161 denotes the banach space of bounded real - linear maps @xmath1162 that are @xmath202-equivariant .
since @xmath1163 is fredholm , we can choose a splitting @xmath1164 such that @xmath1165 is a closed subspace and @xmath1166 maps @xmath1167 isomorphically to its image .
by proposition [ prop : formaladjoint ] , we can similarly split @xmath1168 where @xmath1169 is equivalently the space of all sections in @xmath1170 that are @xmath628-orthogonal to @xmath1171 . in terms of these splittings
, @xmath1172 can be written in block form @xmath1173 where after shrinking @xmath1174 if necessary , we can assume without loss of generality that @xmath1175 is invertible for all @xmath1156 .
we can therefore define a map @xmath1176 [ lemma : zeroset ] a parameter @xmath1134 belongs to @xmath1114 if and only if @xmath1177 .
define for each @xmath1134 the banach space isomorphism @xmath1178 then @xmath1179 , and since @xmath1180 is invertible , @xmath1181 the latter can only have the same dimension as @xmath1182 if @xmath1183 vanishes .
observe that by lemma [ lemma : ndimension ] , corollary [ cor : codimension ] and schur s lemma , @xmath1184 where @xmath342 for representations @xmath343 of real type , complex type and quaternionic type respectively .
the lemma implies via the implicit function theorem that a neighborhood of @xmath1136 in @xmath1114 is a smooth submanifold with the same codimension that appears in theorem [ thm : submanifolds0 ] whenever we can show that the linearization @xmath1185 is surjective .
we will need a precise formula for this linearization in the following special case .
suppose we have a smooth path @xmath1186 such that for all @xmath1187 : 1 .
@xmath1188 ( i.e. there is a canonical complex bundle isomorphism ) ; 2 .
@xmath1189 ; 3 .
@xmath1190 . we are then free to choose the bundle isomorphisms @xmath1154 and consequently the banach space isomorphisms to be the identity for all @xmath1187 , so @xmath1191 , where @xmath1192 is a smooth family of cauchy - riemann operators on the fixed bundle @xmath1193 .
differentiating this family gives a real - linear bundle map @xmath1194 and we then find that @xmath1195 takes the form @xmath1196 where @xmath1197 is the projection @xmath1198 the unique continuation result developed in [ sec : unique ] below is geared toward proving that operators such as @xmath1199 are surjective .
the goal of this section is to compute the fredholm index of the twisted cauchy - riemann type operators introduced in [ sec : twisted ] .
we will use the notation of [ sec : prep ] but dispense with the parameter @xmath402 since it is not important for the index computation , hence @xmath178 is a fixed branched cover , and @xmath239 is a fixed regular presentation .
the complex vector bundles @xmath608 and @xmath1200 with their restrictions @xmath1201 and @xmath1202 to the punctured domain @xmath218 are assumed to have rank @xmath1203 and we assume @xmath927 is a real ( not necessarily irreducible or faithful ) representation of @xmath202 with @xmath1204 the resulting twisted bundles over @xmath218 can be written as @xmath1205 in terms of the flat real vector bundle @xmath1206 , and any cauchy - riemann type operator @xmath615 then gives rise to the twisted operator @xmath1207 we need a bit more notation in order to state a formula for @xmath1208 . recall that while the deck transformations @xmath1209 act on @xmath240 without fixed points , their extensions to biholomorphic self - maps of @xmath566 may fix some of the punctures , so for each @xmath490 and @xmath1210 , we can consider the stabilizer subgroup @xmath1211 which is necessarily cyclic .
restricting @xmath283 to @xmath506 then defines a representation @xmath1212 , which splits @xmath651 into @xmath506-invariant subspaces @xmath1213 such that @xmath506 acts on @xmath1214 trivially and on @xmath1215 as a direct sum of nontrivial representations .
we define the number @xmath1216 as implied by the notation , this depends on @xmath490 but not on the choice of preimage @xmath1217 : indeed , since @xmath202 acts transitively on @xmath1218 , any two choices of @xmath492 give rise to conjugate subgroups @xmath506 , and using orthonormality of characters , one can compute @xmath1219 an expression which depends only on the congugacy class of @xmath506 .
[ thm : index ] under the assumptions detailed above , the operator @xmath1220 is fredholm for any @xmath170 , @xmath171 and negative exponential weights @xmath972 satisfying @xmath1221 for all @xmath490 .
its index is @xmath1222 [ cor : faithful ] if @xmath239 is the mimimal regular presentation and @xmath283 is faithful , then @xmath1223 and the inequality is strict unless all branch points of @xmath26 have branching order @xmath195 . by lemma [ lemma : branchpoints ] ,
the stabilizer subgroups @xmath506 are nontrivial for all @xmath1224 , and the conclusion about branch points of order @xmath195 will hold if and only if all of them are isomorphic to @xmath209 . now
if @xmath283 is faithful , it follows that all nontrivial elements @xmath499 for @xmath1224 also act nontrivially on @xmath651 , hence @xmath1225 for all @xmath490 .
this implies the upper bound , and it is an equality if and only if @xmath1226 for all @xmath490 , meaning each @xmath506 acts on @xmath651 as the @xmath1227-fold direct sum of the trivial representation plus a @xmath53-dimensional nontrivial representation , which is required to be faithful .
but the only nontrivial faithful real @xmath53-dimensional representation of any finite group is the nontrivial representation of @xmath209 , hence @xmath1228 .
the remainder of this section is devoted to the proof of theorem [ thm : index ] .
it will be convenient first to complexify the representation .
we define @xmath1229 and the complex representation @xmath1230 such that @xmath1231 for all @xmath945 .
note that for @xmath490 and @xmath1232 , the trivial representation of @xmath506 on @xmath865 is the complexification of the trivial real representation on @xmath651 , so the splitting @xmath1233 explained above complexifies to a splitting @xmath1234 , where @xmath1235 is the largest complex subspace on which @xmath506 acts trivially , allowing us to write @xmath1236 the complexified representation now gives rise to a complex flat bundle @xmath1237 , corresponding twisted bundles @xmath1238 and a twisted cauchy - riemann operator @xmath1239 the following point is important to understand : the tensor products in are _ real _ , thus @xmath1240 and @xmath1241 each inherit two complex structures @xmath4 and @xmath265 , where @xmath4 comes from the complex structure of @xmath608 and @xmath265 from that of @xmath865 : they commute with each other and are defined by @xmath1242 in this sense , @xmath1243 can be regarded as the @xmath265-complex - linear extension of @xmath1032 to complexifications of the latter s domain and target space this notion of `` complexification '' ignores the fact that these spaces already have native complex structures @xmath4 and treats them as _ real _ vector spaces , which is appropriate since @xmath1032 is not @xmath4-complex linear .
we therefore obtain the relation @xmath1244 and we shall compute @xmath1245 by regarding @xmath1243 as a real - linear cauchy - riemann type operator on the complex vector bundle @xmath1246 . since @xmath1247 , the punctured riemann - roch formula from @xcite gives @xmath1248 where @xmath681 is an arbitrary choice of asymptotic trivialization , and @xmath1249 are conley - zehnder indices that depend on certain asymptotic operators @xmath1250 to be discussed below and the exponential weight @xmath1251 associated to each puncture @xmath490 .
the main difficulty of the calculation is in choosing a suitable asymptotic trivialization in which both @xmath1252 and @xmath1253 can be computed .
denote @xmath1254 and for each @xmath490 and @xmath1232 , let @xmath1255 denote the branching order of @xmath472 at @xmath492 , meaning @xmath472 is a @xmath1256-to-@xmath53 map on a small punctured neighborhood of @xmath492 .
we can then choose punctured neighborhoods @xmath494 and @xmath495 of @xmath453 and @xmath492 respectively , with holomorphic cylindrical coordinates @xmath434 on each such that @xmath1257 in coordinates on @xmath503 .
in these coordinates , any @xmath499 necessarily preserves the end @xmath503 and takes the form @xmath1258 for some @xmath1259 .
this means that @xmath506 is a cyclic group of order @xmath1256 , and it has a canonical generator @xmath1260 such that @xmath1261 in addition to the cylindrical coordinates , let us choose complex trivializations of @xmath608 on each of the corresponding neighborhoods of @xmath238 , thus giving an identification @xmath1262 for each @xmath490 . for any choice @xmath1263 ,
this also gives us an identification of @xmath1264 with @xmath1265 where the action of @xmath1266 on @xmath1267 is determined by @xmath1268 this picture can now easily be extended to the `` circle compactification '' of the punctured surface : let @xmath1269 and @xmath1270 denote the compact surfaces with boundary obtained by replacing each cylindrical end @xmath439 in @xmath218 and @xmath240 respectively by the compact topological manifold @xmath1271
\times s^1 $ ] .
the connected components of @xmath1272 and @xmath1273 are then in bijective correspondence with the punctures @xmath490 or @xmath1224 respectively , and the choice of cylindrical coordinates identifies each of these components with @xmath1274 .
we shall denote the boundary components accordingly by @xmath1275 for @xmath490 or @xmath1224 , hence @xmath1276 the covering map @xmath260 now extends to a continuous covering map @xmath1277 which restricts on the boundary components to @xmath1278 and each @xmath945 also extends naturally to a continuous deck transformation @xmath1279 of @xmath1280 , such that if @xmath1281 , then @xmath1282 maps @xmath1283 via the canonical diffeomorphism composed with a translation .
the identifications and then yield obvious extensions of @xmath1201 and @xmath1240 as topological vector bundles @xmath1284 and we have @xmath1285 we are now prepared to discuss asymptotic operators . recall that after choosing a suitable hermitian inner product on @xmath1201 over the cylindrical ends , any cauchy - riemann type operator @xmath1286 on @xmath1287 with reasonable asymptotic behavior determines real - linear operators @xmath1288 for each @xmath490 , see e.g. @xcite*2.1 .
these can be regarded as unbounded self - adjoint operators on @xmath1289 with dense domain @xmath1290 , and we say @xmath1291 is nondegenerate whenever its kernel is trivial , in which case it determines a conley - zehnder index @xmath1292 relative to any choice of complex trivialization @xmath681 of @xmath1293 . in the case where @xmath1286 is the restriction to @xmath218 of some operator @xmath615 on @xmath192 , the operators @xmath1291 are very simple and were already computed in
[ sec : crpunctured ] : they are each the so - called _ trivial _ asymptotic operator @xmath1294 where @xmath1295 is a well - defined differential operator on @xmath1293 since the fibers are all canonically identified with @xmath1296 . this operator is degenerate , but the introduction of negative exponential weights @xmath1297 identifies @xmath1286 with another cauchy - riemann type operator whose corresponding asymptotic operators are @xmath1298 , which are nondegenerate for any @xmath689 sufficiently small .
denote by @xmath1299 the asymptotic operators associated to @xmath1243 for each @xmath490 .
these are easiest to understand by considering the pulled back cauchy - riemann operator @xmath1300 whose asymptotic operators we will denote by @xmath1301 for @xmath1224 .
the relation @xmath1302 for sections @xmath1303 gives rise to the following relation between asymptotic operators : @xmath1304 for the following discussion , fix @xmath490 and @xmath1217 .
the definition of @xmath1243 implies that @xmath1305 acts on sections @xmath1306 such that @xmath1307 \otimes v$ ] whenever @xmath1308 is constant . from this , one deduces that for any section @xmath1309 where @xmath833 is an arbitrary smooth map @xmath1310 and @xmath1311 is constant , we have @xmath1312 now to write down a formula for @xmath1250 , we can use the natural identification of @xmath1313 with the space of @xmath506-equivariant loops in @xmath1314 , @xmath1315 acting on @xmath506-equivariant loops @xmath1200 , and imply @xmath1316 where it is understood that @xmath1317 acts on the tensor product by taking @xmath1318 to @xmath1319 whenever @xmath165 is locally constant . in order to choose a suitable trivialization @xmath681 and compute @xmath1253 ,
we shall split @xmath1250 into a direct sum of operators on @xmath4-complex line bundles .
first , we observe that @xmath1320 is already canonically trivial , so any complex basis of @xmath1296 gives a splitting of @xmath1250 over an @xmath328-fold direct sum of isomorphic @xmath4-complex bundles of rank @xmath48 , @xmath1321 where @xmath1322 and the generator of @xmath1266 acts by @xmath1323 .
note that @xmath1324 carries two commuting complex structures , @xmath4 and @xmath265 , which act on the first and second factor of the tensor product respectively .
further : @xmath865 admits a complex basis @xmath1325 consisting of eigenvectors of @xmath1326 , and we can then define integers @xmath1327 for @xmath1328 by @xmath1329 here we can identify @xmath1330 as the subspace spanned by all @xmath1331 such that @xmath1332 . identifying @xmath865 with @xmath1333 via this
eigenbasis yields a splitting @xmath1334 where for @xmath1328 , @xmath1335 with the generator @xmath1336 acting by @xmath1337 .
this bundle again carries the two commuting complex structures @xmath4 and @xmath265 acting on the first and second factors of the tensor product respectively ; it has complex rank @xmath195 with respect to either one . finally , since @xmath4 acts @xmath265-complex - linearly on @xmath1338
, we can find eigenvectors @xmath1339 such that @xmath1340 , so the splitting @xmath1341 gives a splitting of @xmath4- and @xmath265-complex vector bundles @xmath1342 with @xmath1343 where the generator @xmath1336 acts by @xmath1344 .
both @xmath1345 and @xmath1346 are complex line bundles over @xmath1274 , carrying two complex structures @xmath4 and @xmath265 , which satisfy @xmath1347 on @xmath1345 but @xmath1348 on @xmath1346 .
this splitting of bundles gives a splitting of @xmath1250 in the form @xmath1349 where for @xmath1328 , @xmath1350 acts on @xmath1351 by @xmath1352 since @xmath1353
are complex line bundles , @xmath1354 can be computed in terms of winding numbers of eigenfunctions of @xmath1350 , using the relation proved in @xcite*theorem 3.10 . in particular ,
if ( as will turn out to be true in our case ) all eigenspaces of @xmath1350 have real dimension @xmath195 , then @xmath1355 where @xmath1356 is any nontrivial eigenfunction of @xmath1357 with the largest possible negative eigenvalue .
a @xmath1358-equivariant function @xmath1359 satisfies @xmath1360
if and only if it is a complex multiple of @xmath1361 observe that since @xmath1362 , every eigenvalue @xmath1363 thus satisfies @xmath1364 ; this proves that the perturbed asymptotic operators @xmath1350 are all nondegenerate and thus establishes the fredholm property for @xmath1243 . now to apply , we need to find the unique eigenvalue @xmath1365 for @xmath1366 such that @xmath1367 - \delta_w.\ ] ] since @xmath1368 , this condition is equivalent to @xmath1369 so choosing the appropriate @xmath1366 and plugging in leads to the formulas @xmath1370 let @xmath1371 for @xmath1328 denote a choice of @xmath4-complex trivializations of @xmath1353 such that @xmath1372 and denote by @xmath1373 the resulting @xmath4-complex trivialization of @xmath1374 by , we now have @xmath1375 and thus by , @xmath1376 note that , a priori , this construction of @xmath1373 depends on an arbitrary choice @xmath1217 , but the fact that @xmath1377 turns out to be independent of this choice tells us that @xmath1373 is uniquely determined up to homotopy . performing this construction for all punctures @xmath490 , we will denote the resulting asymptotic trivialization of @xmath1240 simply by @xmath681 .
it remains to compute @xmath1252 .
consider the pullback @xmath1378 .
the first factor in this tensor product has a canonical asymptotic trivialization , which we shall denote by @xmath1379 , as it is the pullback of an asymptotic trivialization @xmath1380 for @xmath1201 , satisfying @xmath1381 .
moreover , the second factor is globally trivial , thus @xmath1382 carries a canonical asymptotic trivialization , denoted by @xmath1383 , such that @xmath1384 if @xmath1385 denotes the pullback of @xmath681 to an asymptotic trivialization of @xmath1382 , we then have @xmath1386 \\ & = 2n \cdot c_1(e ) + \frac{1}{d ' } \deg^{\psi}(\pi^*\phi ) , \end{split}\ ] ] where @xmath1387 denotes the sum over all punctures @xmath1224 of the degrees of the transition maps @xmath1388 that change @xmath1383 to @xmath1385 .
we can compute the latter for each @xmath490 and @xmath1232 as a sum of winding numbers over a line bundle decomposition analogous to , namely @xmath1389 where pulling back via the projection @xmath1390 gives the trivial line bundle @xmath1391 with the pulled back trivialization @xmath1392 such that the special eigenfunctions @xmath1393 in have zero winding as @xmath839 traverses @xmath1274 .
the restriction @xmath1394 of @xmath1383 to @xmath1395 is now the direct sum of the standard trivializations on each of the factors @xmath1396 , thus @xmath1397.\ ] ] there is an important sublety here : recall that @xmath1398 on @xmath1353 , hence the orientation induced by @xmath4 on @xmath1346 is the _ opposite _ of the obvious one , and the sign of @xmath1399 must be reversed accordingly , giving @xmath1400 plugging this into , we have @xmath1401 summing over all @xmath1224 and plugging into then gives @xmath1402 where we ve used the fact that @xmath1403 is independent of @xmath1217 for each @xmath490 , and @xmath1404 .
finally , we combine the above expression with the conley - zehnder index and plug into to obtain @xmath1405 and thus @xmath1406 - m \sum_{w \in \theta } n_w.\ ] ] the expression in brackets is @xmath1407 , so this completes the proof of theorem [ thm : index ] .
standard proofs of transversality results via the sard - smale theorem ( cf .
@xcite ) typically require some kind of unique continuation lemma , which for @xmath4-holomorphic curves usually means the similarity principle .
a less standard result of this type is also needed for the proof of theorem [ thm : submanifolds0 ] , and is the topic of the present section .
the following is a local result , so we do not need to make any global assumptions about the riemann surface @xmath192 or vector bundles @xmath608 and @xmath1200 . in analogy with cauchy - riemann type operators on a vector bundle @xmath1408 over a riemann surface , we shall refer to a real - linear first - order partial differential operator @xmath1409 as an anti - cauchy - riemann type operator of it satisfies the leibniz rule @xmath1410 where @xmath1411 .
two classes of examples immediately come to mind : first if @xmath635 is a cauchy - riemann type operator on @xmath608 and we denote by @xmath1412 the same real bundle with its complex structure modified by a sign change , then @xmath635 defines an anti - cauchy - riemann type operator on @xmath1412 .
more importantly for our purposes , the formal adjoint @xmath664 can be regarded as an anti - cauchy - riemann operator on @xmath1413 after using the relevant bundle metrics to identify @xmath630 with @xmath1414 .
we say that a smooth section of a vector bundle @xmath1408 vanishes to infinite order at a point @xmath1415 if in some choice of local trivialization near @xmath432 , its derivatives of all orders vanish at @xmath432 .
a section that does not vanish to all orders at any point has the property that it is nonzero on an open and dense subset of @xmath192 .
the standard unique continuation result for cauchy - riemann type equations states that nontrivial solutions to such equations do not vanish to infinite order at any point .
note that by complex conjugation , the same is true of solutions to anti - cauchy - riemann type equations .
[ prop : unique ] suppose @xmath608 and @xmath1200 are smooth finite - rank complex vector bundles over a riemann surface @xmath192 , with a cauchy - riemann type operator @xmath1416 and an anti - cauchy - riemann type operator @xmath1417 .
then for any sets of complex - linearly independent sections @xmath1418 no linear combination @xmath1419 with coefficients @xmath1420 satisfying @xmath1421 for some @xmath1422 vanishes to infinite order at any point .
recall from lemma [ lemma : notcomplex ] that every cauchy - riemann type operator can be perturbed to one whose kernel and cokernel are _ totally real _ , meaning they have real bases that are also complex - linearly independent . [ cor : unique ] under the assumptions of proposition [ prop : unique ] , suppose @xmath1423 and @xmath1424 are finite - dimensional totally real subspaces of @xmath630 and @xmath631 respectively .
then the natural map @xmath1425 defined by @xmath1426 is injective , and the nontrivial sections in its image do not vanish to infinite order at any point .
denote the complex structures on @xmath608 and @xmath1200 by @xmath1427 and @xmath1428 respectively .
the proposition implies that the natural map @xmath1429 has the stated property , but this map factors through @xmath1430 and the bundle map @xmath1431 defined via the equivalence relation @xmath1432 . before proving proposition [ prop : unique ] ,
let us sketch why it is true in the simplest case , where @xmath635 and @xmath1433 are both complex linear .
we can then write these as @xmath1434 and @xmath1435 respectively in some local trivialization , so we are left with proving that for any complex - linearly independent sets of holomorphic functions @xmath1436 and antiholomorphic functions @xmath1437 , nontrivial linear combinations @xmath1438 can not vanish to infinite order at @xmath0 .
the key is to look at the taylor series for @xmath1439 and @xmath1440 : write @xmath1441 for @xmath1442 and @xmath1443 , and note that linear independence means there do not exist any constants @xmath1444 such that @xmath1445 for all @xmath59 or @xmath1446 for all @xmath510 except the trivial case with all @xmath1447 .
we then have @xmath1448 which vanishs to infinite order at @xmath0 if and only if @xmath1449 for every @xmath1450 .
the linear independence of @xmath1451 then implies @xmath1452 for all @xmath265 and @xmath59 , so that linear independence of @xmath1453 implies in turn that @xmath1454 for all @xmath1422 .
the above argument does not work in the general case since the taylor series of @xmath1439 and @xmath1440 at @xmath0 can not be assumed to be purely holomorphic and antiholomorphic respectively .
we shall therefore modify the argument to use only information about the _
first nontrivial term _ in such taylor expansions .
note that in suitable choices of local coordinates and trivializations , the operators @xmath635 and @xmath1433 in the lemma can always be identified with @xmath1455 where @xmath1456 and @xmath1457 are smooth functions on the disk @xmath431 valued in @xmath1458 and @xmath1459 respectively .
[ lemma : firstterm ] if @xmath700 and @xmath1460 are as described above , then for any nontrivial solutions @xmath1461 and @xmath1462 to the equations @xmath1463 and @xmath1464 , the first nontrivial terms in the taylor series for @xmath662 and @xmath1465 at @xmath0 are holomorphic and antiholomorphic respectively .
suppose @xmath1466 and @xmath1467 for all multiindices @xmath760 of order @xmath1468 but not for all @xmath1469 .
then if @xmath1470 , we have @xmath1471 leaving @xmath1472 as the only nonzero derivative of order @xmath1473 . a similar argument using
the equation @xmath1474 proves that if @xmath1465 has vanishing derivatives at @xmath0 up to order @xmath59 but not @xmath1473 , then its only nonzero derivative of order @xmath1473 is @xmath1475 .
choosing suitable local coordinates and trivializations , we shall assume the @xmath1439 and @xmath1440 are functions @xmath1476 and @xmath1477 respectively , with @xmath635 and @xmath1433 written as in .
suppose there exist constants @xmath1420 such that @xmath1478 vanishes to infinite order at @xmath0 .
we then claim that @xmath1479 for all multiindices @xmath760 and @xmath1480 .
we shall prove this by induction on the total order @xmath1481 of the multiindices .
the claim clearly holds for @xmath1482 , so for a given @xmath170 , assume it holds for all pairs of multiindices satisfying @xmath1483 .
denote the individual complex components of @xmath1436 and @xmath1437 by @xmath1484 and @xmath1485 for @xmath1486 , @xmath1487 .
then is equivalent to the relation @xmath1488 for all @xmath1489 .
if this is true whenever @xmath1483 , it implies @xmath1490 and @xmath1491 again for all @xmath1492 whenever @xmath1483 .
observe that @xmath1493 and @xmath1494 are each linear combinations of functions in @xmath1495 and @xmath1496 respectively and thus also belong to these kernels .
fixing a multiindex @xmath760 with @xmath1497 , implies that @xmath1493 vanishes up to order @xmath1498 at @xmath0 , so lemma [ lemma : firstterm ] implies that @xmath1499 also holds for all @xmath1500 with the possible exception of the holomorphic taylor coefficient @xmath1501 .
similarly , if @xmath1480 is a fixed multiindex with @xmath1502 , then implies that @xmath1494 vanishes up to order @xmath1498 at @xmath0 , and lemma [ lemma : firstterm ] extends this to order @xmath1503 with the possible exception of the antiholomorphic coefficient @xmath1504 .
this proves all cases of with @xmath1505 except those for which @xmath1506 is a power of @xmath1434 and @xmath1507 is a power of @xmath1435 .
now if @xmath1508 is any multiindex with @xmath1509 , the assumption that @xmath1478 vanishes to infinite order at @xmath0 implies @xmath1510 for suitable combinatorial constants @xmath1511 , and this expansion contains exactly one term that has not already been shown to vanish .
this implies all remaining cases of with @xmath1505 and thus completes the inductive argument .
the claim implies that the functions @xmath1493 and @xmath1494 defined above all vanish to infinite order at @xmath0 , and since they satisfy @xmath1512 and @xmath1513 respectively , the usual unique continuation results imply that they all vanish identically . since the sets @xmath1453 and @xmath1451 are linearly independent , then implies @xmath1514 while implies @xmath1515 since @xmath1516 and @xmath1517 each satisfy cauchy - riemann or anti - cauchy - riemann type equations , we can again appeal to unique continuation and conclude that both are identically zero , which contradicts the linear independence assumption unless all @xmath1454 .
we now give an application of corollary [ cor : unique ] which will be crucial for the proof of theorem [ thm : submanifolds0 ] .
the setting is as in [ sec : crpunctured ] : assume @xmath39 is a closed riemann surface with a finite set of points @xmath113 , denote by @xmath218 the punctured surface @xmath119 , assume @xmath1408 is a complex vector bundle of rank @xmath1518 , @xmath1413 , and @xmath1519 denote the restrictions of @xmath608 and @xmath1200 to the punctured surface .
fixing holomorphic cylindrical coordinates near the punctures and a cauchy - riemann type operator @xmath615 , its restriction to the punctured domain defines a bounded linear map of weighted sobolev spaces @xmath1520 for all @xmath170 , @xmath171 and exponential weights @xmath1521 , which is fredholm if all @xmath1522 are chosen to be positive and sufficiently small .
fixing suitable bundle metrics , the formal adjoint @xmath1523 is then also fredholm and is an anti - cauchy - riemann type operator on @xmath1202 such that by proposition [ prop : formaladjoint ] , @xmath1524 allowing us to identify @xmath1525 with the @xmath628-orthogonal complement @xmath1526 of @xmath1527 for @xmath739 .
let @xmath1528 denote the resulting projection along @xmath1529 .
recall from definition [ defn : notcomplex ] the open and dense set @xmath646 of cauchy - riemann type operators whose kernels and cokernels are guaranteed to be totally real .
[ lemma : surjective ] for any open subset @xmath1530 , assume @xmath1531 is a linear subspace satisfying the following conditions : 1 .
all @xmath1532 have compact support in @xmath35 .
2 . for every @xmath1533 and @xmath1534
, there exists a fixed @xmath1535 satisfying @xmath1536 such that for every neighborhood @xmath1537 of @xmath432 , there also exists a smooth compactly supported function @xmath1538 $ ] with @xmath1539 and @xmath1540 .
then for any @xmath649 , the linear map @xmath1541 defined by @xmath1542 is surjective .
fix bases @xmath1543 and @xmath1544 , which are each complex - linearly independent since @xmath1545 . note that @xmath628-products of sections in @xmath1546 with elements of @xmath1547 are always well defined since @xmath1548 for @xmath739 by lemma [ lemma : regularity2 ] ; in particular , the @xmath628-product has a well - defined restriction to @xmath1547 itself . by the orthogonality of @xmath1549 to @xmath1547
, we then have @xmath1550 and these matrix elements determine @xmath1551 . now
if @xmath1199 is not surjective , there exists a nontrivial linear map @xmath1552 which is `` orthogonal '' to every @xmath1553 in the sense that its matrix elements @xmath1554 satisfy @xmath1555 for every @xmath1532 .
denoting the measure on @xmath218 by @xmath1556 and the hermitian inner product on @xmath1200 by @xmath1557 , we can rewrite as @xmath1558 where @xmath1559 is regarded as a section of @xmath1560 .
since the @xmath1561 are not all zero by assumption , this section is nonzero on an open and dense subset by corollary [ cor : unique ] . choosing a point @xmath1533 where it is nonzero
, lemma [ lemma : tensor ] below then provides a linear map @xmath1562 such that the integrand is positive near @xmath432 for any @xmath1532 satisfying @xmath1536 , and we can then make the entire integral positive after multiplying @xmath681 by smooth bump functions with sufficiently small support .
we used : [ lemma : tensor ] suppose @xmath865 and @xmath651 are real finite - dimensional vector spaces , @xmath1563 is an inner product on @xmath865 , and @xmath1564 is nonzero . then there exists a linear map @xmath1565 such that @xmath1566 . choosing a basis @xmath1567 of @xmath651 , we have @xmath1568 for unique vectors
@xmath1569 that do not all vanish since @xmath1570 . choosing @xmath1565 such that @xmath1571 for all @xmath99 then gives @xmath1572 .
we are now in a position to prove theorem [ thm : submanifolds0 ] .
the main idea behind the proof is standard , though some details are less so : we will define a suitable banach manifold of perturbed almost complex structures , producing a universal moduli space and a projection to the space of perturbed data whose regular values have the property stated in the theorem .
the hard part is of course to prove that the universal moduli space is a smooth banach manifold
this follows from the implicit function theorem after proving that some version of the operator defined in is surjective , and that is where lemma [ lemma : surjective ] from the previous section is needed .
we start by using the floer @xmath1573-topology ( cf .
@xcite*5 ) to define spaces of perturbed data .
fix a sequence of positive numbers @xmath1574 and , given a closed riemann surface @xmath39 with complex vector bundles @xmath1408 and @xmath1413 , define the separable banach space @xmath1575 where @xmath1576 if @xmath1577 , we can view the `` tangent space '' @xmath1578 as the space of smooth @xmath1579-antilinear bundle maps @xmath1580 that vanish outside @xmath35 and satisfy @xmath1581 , and there is a natural embedding @xmath1582 which takes a @xmath1583-small neighborhood of @xmath0 in @xmath1578 homeomorphically to a neighborhood of @xmath1579 in @xmath37 .
this allows us to define for any @xmath1584 sufficiently small a smooth , separable and metrizable banach manifold @xmath1585 which embeds continuously into @xmath1586 and contains arbitrarily @xmath36-small perturbations of @xmath1579 .
recall that a point @xmath1415 in the domain of a smooth map @xmath1587 is called an injective point if @xmath1588 is injective and @xmath1589 . for a simple @xmath4-holomorphic curve
, the complement of the set of injective points is a discrete set .
[ lemma : operatorpert ] assume @xmath1590 , @xmath177 is a simple @xmath4-holomorphic curve with generalized normal bundle @xmath1591 and normal cauchy - riemann operator @xmath1592 , and @xmath1593 has support contained in the set of injective points in @xmath1594
. then there exists a smooth family of almost complex structures @xmath1595 such that @xmath1596 , @xmath1597 for all @xmath839 and @xmath432 , and the resulting family of normal cauchy - riemann operators @xmath1598 for @xmath165 defined with respect to @xmath1599 satisfies @xmath1600 if @xmath1601 is any smooth path in @xmath556 with @xmath1596 , @xmath1602 for all @xmath839 and @xmath1603 , then @xmath1604 , hence @xmath1605 is well defined along @xmath165 independently of any connection . for @xmath1606 ,
let us write @xmath1607 in block form as @xmath1608 with respect to the tangent - normal decomposition @xmath382 .
choosing @xmath1609 for convenience to be the @xmath30-symplectic orthogonal complement of @xmath1610 , the fact that @xmath1599 is always @xmath30-compatible then translates into conditions that constrain @xmath1611 and @xmath1612 separately and another condition that determines @xmath1613 in terms of @xmath1614 , namely @xmath1615 for all @xmath1616 .
this means that @xmath30-compatibility does not prevent us from freely choosing @xmath1614 so long as we ( 1 ) do not mind @xmath1613 being determined by this choice , and ( 2 ) do this only in regions where @xmath165 has no double points , so that the splitting of @xmath1617 into @xmath1618 is unambiguous . now using the definition of the normal cauchy - riemann operator
, one computes that for any @xmath1606 , @xmath1619 on a region where @xmath165 has neither critical points nor double points and its image lies in the perturbation domain @xmath35 , we can therefore choose the normal derivatives of @xmath1620 along @xmath165 to make the above expression match @xmath700 .
we recall next from definition [ defn : notcomplex ] the space of cauchy - riemann operators whose complex - antilinear parts are invertible on some fiber .
[ lemma : notcomplex2 ] there exists a baire subset @xmath1621 such that for all @xmath1622 and all @xmath109 , @xmath320 and @xmath110 , every simple curve @xmath1623 passing through @xmath35 has @xmath1624 .
abbreviate @xmath1625 and define @xmath1626 as the space of curves @xmath177 with marked points @xmath1627 such that @xmath1628 is the set of critical points of @xmath165 , satisfying @xmath1629 for @xmath323 , and @xmath1630 is an injective point with @xmath1631 . for each pair of integers @xmath1632 , we then define a subspace @xmath1633 via the following constraint . abbreviating the complex vector bundles @xmath1634 over @xmath192 ,
define the fiber bundle @xmath1635 whose fiber over @xmath1415 is the submanifold of complex - antilinear maps @xmath1636 with @xmath1637 .
these fibers have complex codimension @xmath1638 in @xmath1639 , hence @xmath1640 is a submanifold of @xmath1641 with real codimension @xmath1642 .
the curve @xmath165 determines a natural section of @xmath1641 , namely the antilinear part of the normal cauchy - riemann operator , @xmath1643 and we define @xmath165 to be an element of @xmath1644 if and only if this section meets @xmath1640 at @xmath1630 with a tangency of order @xmath1645 , i.e. its @xmath1645-jet matches that of a map with image in @xmath1640 .
the two spaces defined above have corresponding universal moduli spaces @xmath1646 each consisting of pairs @xmath1647 where @xmath1590 and @xmath165 belongs to either @xmath1648 or @xmath1644 respectively .
the arguments of appendix [ sec : critical ] show that @xmath1649 is a smooth , metrizable and separable banach manifold ; indeed , for any element @xmath1650 represented by @xmath540 , one can identify a neighborhood of @xmath557 in @xmath1649 with some finite - codimensional smooth submanifold @xmath1651 in the smooth zero - set of the nonlinear cauchy - riemann operator @xmath1434 defined on the product of @xmath556 with @xmath549 for @xmath171 and @xmath170 sufficiently large , and a teichmller slice @xmath550 through @xmath551 ( see appendix [ sec : critical ] ) .
the submanifold @xmath1652 is determined by constraints on the jet evaluation map that produce the correct critical orders for the first @xmath328 marked points ; the only difference between this and the setup in the appendix is that since there is one extra marked point , the teichmller slice is larger by @xmath195 dimensions .
now we claim that the intersection of this neighborhood of @xmath557 with @xmath1653 defines a smooth submanifold of @xmath546 with real codimension @xmath1654 indeed , the condition defining @xmath1653 constrains the @xmath1645-jet of @xmath1655 at @xmath1630 to intersect a submanifold of precisely this dimension , i.e. the product of @xmath1656 with @xmath1657 , which is the number of distinct multi - indices up to order @xmath1645 for a function of two real variables .
this intersection is also transverse : to see this , observe that by lemma [ lemma : operatorpert ] , @xmath1658 contains elements of the form @xmath1659 , where @xmath1660 can be chosen to produce arbitrary perturbations to @xmath191 of class @xmath1573 near @xmath1661 . in particular , @xmath1620 can be chosen to perturb @xmath1655 by a section with any desired taylor polynomial of some finite but large degree at @xmath1630 . with the claim established
, we can apply the sard - smale theorem to the projection @xmath1662 and thus find a baire subset of @xmath556 for which @xmath1648 and @xmath1644 are both smooth finite - dimensional manifolds and the codimension of the latter in the former is given by .
this means @xmath1663 will always become negative when @xmath1645 is chosen large enough , and will therefore be empty for @xmath4 in our baire subset .
finally , observe that any simple curve @xmath1623 that passes through @xmath35 and has @xmath1664 can be regarded as an element of the space @xmath1644 for some @xmath1665 and arbitrarily large @xmath1666 after adding an extra marked point somewhere in @xmath1594 .
the proof above would work equally well to find generic families of almost complex structures depending on finitely many parameters , since the codimension in can be made arbitrarily large by choosing @xmath1645 larger .
this detail is important for the bifurcation theory discussed in [ sec : bifurcations ] .
we now proceed toward the proof of theorem [ thm : submanifolds0 ] . for each of the choices of data in the statement of theorem [ thm : submanifolds0 ] , we define two universal moduli spaces @xmath1667 which consist of pairs @xmath1668 with @xmath1590 , where in the first case @xmath7 is assumed to belong to @xmath330 , and in the second @xmath1669 .
writing @xmath348 for the underlying simple curve @xmath165 , we further impose the open condition @xmath1670 on both of these spaces , so that the kernel and cokernel of any pullback of @xmath191 are guaranteed by lemma [ lemma : notcomplex ] to be totally real .
the first space is easily seen to be a smooth separable banach manifold : indeed , suppose @xmath1671 , where @xmath1672 for some simple @xmath542-holomorphic curve @xmath540 and @xmath44-fold branched cover @xmath1673 representing an element of @xmath1674 with generalized automorphism group @xmath202 .
then @xmath557 lives in the smooth universal moduli space @xmath1675 defined in appendix [ sec : critical ] , more specifically in the finite - codimensional submanifold @xmath1676 of this space defined by the condition that the @xmath265th marked point should have critical order @xmath322 and curves are immersed everywhere else . as in the proof of lemma
[ lemma : notcomplex2 ] , a neighborhood of @xmath557 in @xmath1677 can thus be identified with a smooth finite - codimensional submanifold @xmath1651 of the zero - set of the nonlinear cauchy - riemann operator . in parallel with this
, we can parametrize a neighborhood of @xmath355 in @xmath1674 as explained in examples [ ex : teichmueller ] and [ ex : teichmueller3 ] , meaning that if @xmath1678 is the set of critical values of @xmath355 , we choose a smooth family of diffeomorphisms @xmath518 parametrized by @xmath1679 which are holomorphic near @xmath238 and supported on a slightly larger neighborhood of @xmath238 such that @xmath536 and @xmath537 is a diffeomorphism onto an open set .
the neighborhood of @xmath1680 in @xmath1681 can now be identified with @xmath1682 by associating to each @xmath1683 the curve @xmath1684 .
following example [ ex : teichmueller3 ] , this fits into the general picture from [ sec : prep ] of a parametrized family of bundles with cauchy - riemann operators , where the parameter space is the banach manifold @xmath1685 and for any @xmath1686 we have the data * @xmath555 , * @xmath535 , * @xmath1687 , * @xmath1688 .
we should pause a moment to discuss why the generalized normal bundles @xmath1609 form a smooth family as @xmath1689 varies this depends crucially on the critical point constraints satisfied by @xmath165 .
assuming @xmath1591 is always defined as the symplectic orthogonal complement of @xmath1690 with @xmath1691 away from critical points , let us recall from @xcite how the latter is defined at critical points .
we have a smooth family of bundles @xmath1692 carrying linearized cauchy - riemann operators @xmath1693 , whose complex - linear parts @xmath1694 define a smooth family of holomorphic structures on @xmath1692 .
the crucial observation is then that @xmath1695 is always a holomorphic section with respect to the holomorphic bundle structures on @xmath1692 and @xmath627 , so choosing a smooth family of holomorphic trivializations and holomorphic coordinates near the @xmath265th marked point , each @xmath383 is represented by some holomorphic function of the form @xmath1696 where @xmath1697 is another family of holomorphic functions which depend smoothly on @xmath1689 but also are nonzero at @xmath0 .
the main point here is that the critical orders @xmath322 do not vary with @xmath165 .
the span of @xmath1698 thus defines the fibers of @xmath1610 near each critical point , so we deduce smooth dependence of @xmath1610 on @xmath1689 . if @xmath1699 , then using the setup in [ sec : ift ] , we now find a regular branched cover @xmath1700 with automorphism group @xmath1701 and corresponding @xmath542-holomorphic curve @xmath1702 , with a smooth map @xmath1703 whose zero - set is a neighborhood of @xmath1680 in @xmath1704 . here the normal operator @xmath1705 is defined over the punctured domain @xmath1123 with negative exponential weights close to zero , and its formal adjoint @xmath1706 has correspondingly positive ( but small ) exponential weights .
the linearization of @xmath1707 at @xmath1708 is surjective . by lemma [ lemma : operatorpert ]
, the tangent space to @xmath1682 at @xmath1708 contains a large space of elements of the form @xmath1709 for @xmath1710 which can be chosen to realize any perturbation in @xmath1711 of the form @xmath1712 where @xmath1713 is a zeroth - order term of class @xmath1573 with support in @xmath1714 away from the critical and double points of @xmath545 .
the resulting change in @xmath1715 is @xmath1716 hence differentiating @xmath1707 in the direction @xmath1709 produces a @xmath202-equivariant linear map @xmath1717 given by , namely @xmath1718 in terms of the projection @xmath1719 we claim that @xmath1620 can be chosen to make @xmath1720 equal to any given element @xmath1721 indeed , let us abbreviate @xmath1722 and @xmath1723 . since @xmath1724 and
therefore also @xmath1725 , lemma [ lemma : surjective ] provides for any given @xmath1383 a section @xmath1726 satisfying @xmath1727 , with support in some small neighborhood of a point @xmath1728 where @xmath1729 and @xmath1730 is an injective point of @xmath545 , such that @xmath1731 for all @xmath1732 and @xmath1733 . note that we are free to assume the @xmath628-product is invariant under the action of @xmath202 via deck transformations . then since @xmath1383 is @xmath202-equivariant , we also have for every @xmath945 , @xmath1734 implying that the symmetrization @xmath1735 also satisfies @xmath1736 for all @xmath1737 .
but the @xmath202-invariance of @xmath1738 implies @xmath1739 for some @xmath1740 with @xmath1710 , so we are done .
the lemma implies via the implicit function theorem that @xmath1741 is a smoooth banach submanifold with codimension given by the formula in .
we can then apply the sard - smale theorem to the projection @xmath1742 and thus find a baire subset in @xmath556 for which the statement of theorem [ thm : submanifolds0 ] holds so long as we exclude all covers of simple curves @xmath165 with @xmath1664 .
intersecting further with the baire subset provided by lemma [ lemma : notcomplex2 ] , we can assume without loss of generality that no such curves exist .
since the reference structure @xmath1579 in the definition of @xmath556 was arbitrary , this implies in particular that the theorem holds for a _ dense _ set of @xmath4 in @xmath37 . to turn this into a baire subset
, we now apply the standard taubes trick ( cf .
@xcite*4.4.2 ) : the idea is to exhaust @xmath1669 by a countable sequence of compact subsets @xmath1743 for @xmath1744 , each defined as the set of curves @xmath1745 that lie at distance at least @xmath1746 separated ( with respect to some metric ) from any of the closed subsets in the gromov compactification where @xmath26 becomes a nodal branched cover , or two of its critical values run together , or @xmath165 becomes a nodal curve , or a multiple cover , or its image escapes from @xmath35 , or its algebraic sum of critical points increases .
this can be done in a way that depends continuously on @xmath4 , so the subset @xmath1747 consisting of @xmath4 for which @xmath1748 is a submanifold of the correct dimension ( meaning it is locally the zero - set of a submersion to a vector space of that dimension ) is open , while the sard - smale argument above shows that it is also dense .
the countable intersection of these open and dense subsets is therefore the baire subset of @xmath37 promised by theorem [ thm : submanifolds0 ] , and the proof of the theorem is complete .
we now prove the @xmath1-dimensional case of theorem [ thm : super ] , using intersection - theoretic arguments that are essentially unrelated to the rest of the paper . throughout this section ,
assume @xmath151 is an almost complex manifold with @xmath1749 the genus zero case is an `` automatic '' phenomenon , i.e. it does not require any genericity condition except for ensuring that the index @xmath0 simple curve is immersed : [ prop : genuszero ] every simple immersed @xmath4-holomorphic sphere @xmath1750 of index @xmath0 in an almost complex @xmath1-manifold is super - rigid .
assume @xmath1751 is a @xmath44-fold branched cover and @xmath348 . since @xmath165 is immersed , the riemann - roch formula implies @xmath1752 hence @xmath1753 .
then @xmath1754 , so if @xmath1755 is nontrivial , its algebraic count of zeroes is negative , violating the similarity principle . for the genus one case
, we use a variant of the `` magic trick '' proposed by hutchings @xcite in the context of embedded contact homology .
[ prop : genusone ] a simple immersed @xmath4-holomorphic torus @xmath1756 of index @xmath0 in an almost complex @xmath1-manifold is super - rigid if and only if all its unbranched covers are fredholm regular .
we will assume for most of the proof that @xmath177 has unspecified genus @xmath1757 .
since @xmath165 is immersed with index @xmath0 , it is regular if and only if its normal cauchy - riemann operator @xmath191 is injective , so given this and the assumption that the same holds for all unbranched covers @xmath348 , we need to show that @xmath20 is also injective for @xmath348 where @xmath178 is _ any _ holomorphic branched cover .
we will prove this by induction on the degree @xmath1758 , thus assume it is true for all covers up to degree @xmath87 .
note that since @xmath1759 , we have @xmath1760 and if @xmath26 has branch points , then @xmath1761 by the riemann - hurwitz formula . by the construction in the proof of proposition [
prop : isolated ] , one can endow the total space of the normal bundle @xmath1762 with an almost complex structure @xmath1763 such that @xmath1763-holomorphic curves @xmath1764 correspond to sections @xmath1765 along holomorphic branched covers @xmath1766 .
if @xmath1767 contains a nontrivial element @xmath662 , the inductive hypothesis implies that the corresponding @xmath1763-holomorphic curve @xmath1768 is somewhere injective .
we can view @xmath165 itself as a @xmath1763-holomorphic embedding into @xmath1609 , and @xmath1768 is homologous to its @xmath44-fold cover , so applying the adjunction formula to both @xmath1768 and @xmath165 as @xmath1763-holomorphic curves in @xmath1609 , @xmath1769 where @xmath1770 denotes the algebraic count of double points and critical points of @xmath1768 .
solving for @xmath1771 and plugging in to compute @xmath1772 , we have @xmath1773 plugging in @xmath1774 and the fact that @xmath1761 , this gives a contradiction since @xmath1771 can not be negative .
[ remark : bifurcationsfour ] in the spirit of [ sec : bifurcations ] , the two results above show that the story of super - rigidity and bifurcations is simpler in dimension four . in the genus
zero case bifurcations can be avoided altogether : since having a critical point is a codimension @xmath195 condition ( see appendix [ sec : critical ] ) , index @xmath0 simple curves for generic @xmath53-parameter families of almost complex structures can be assumed immersed , and therefore super - rigid by prop .
[ prop : genuszero ] .
this is no longer true in the genus one case since regularity of some unbranched cover might fail under a generic homotopy , producing the birth - death or degree - doubling bifurcations in @xcite , but prop .
[ prop : genusone ] implies that this is the only danger the only bifurcations that can happen involve unbranched covers with @xmath1775 and @xmath1776 , and they are already described in @xcite .
the proof of prop .
[ prop : genusone ] also gives a partial result for higher genus , though it is not very strong .
suppose @xmath7 is a @xmath44-fold cover with genus @xmath1777 of a simple index @xmath0 curve @xmath165 with genus @xmath1778 that is regular and immersed .
if additionally @xmath1779 then either @xmath20 is injective or @xmath7 factors through another cover @xmath371 of @xmath165 with smaller degree such that @xmath1780 is nontrivial .
the theorem below is well known to experts , but a proof of it is difficult to find in the literature , so we will sketch one here . fix a symplectic manifold @xmath27 of dimension @xmath48 , @xmath1781 , and suppose @xmath1782 .
recall that if @xmath39 is a connected riemann surface and @xmath1783 is a nonconstant @xmath4-holomorphic curve with a critical point @xmath1784 , then the critical point is isolated and has a well - defined positive order , @xmath1785 characterized by the property that @xmath1786 if @xmath432 is a zero of order @xmath510 for @xmath1787 , where the latter is viewed as a holomorphic section with respect to a natural holomorphic bundle structure on @xmath1788 determined by the linearized cauchy - riemann operator , see e.g. @xcite*3.3 .
when @xmath39 is closed , we denote the resulting algebraic count of critical points by @xmath1789 and note that it vanishes if and only if @xmath7 is immersed .
given integers @xmath109 , a homology class @xmath110 and a tuple of positive integers @xmath1790 , let @xmath1791 denote the following subset of the moduli space of unparametrized @xmath4-holomorphic curves homologous to @xmath700 with genus @xmath69 and @xmath328 marked points : a map @xmath3 with marked points @xmath1792 representing an element of @xmath111 belongs to @xmath1793 if and only if it is critical at all marked points , @xmath1794 and it is immersed everywhere else . [ prop : critical ] fix an open subset @xmath33 with compact closure and a compatible almost complex structure @xmath1795 . there exists a baire subset @xmath1796 such that for all @xmath1797 and all @xmath109 , @xmath110 and @xmath1798 , the open subset of @xmath1793 consisting of somewhere injective curves that pass through @xmath35 is a smooth manifold with dimension equal to its virtual dimension , where @xmath1799 for generic compatible @xmath4 in any closed symplectic @xmath48-manifold , all closed , connected and somewhere injective @xmath4-holomorphic curves @xmath7 with @xmath319 critical points satisfy @xmath1800 .
one well - known consequence of this result is that for generic @xmath4 , somewhere injective index @xmath0 curves in almost complex manifolds of dimension at least four are always immersed .
another proof of this is given in @xcite , though it is analytically somewhat more complicated than the one given below .
it will suffice to prove that the same statement as in prop .
[ prop : critical ] holds for the slightly larger moduli space @xmath1801 characterized by the condition @xmath1802 for all @xmath1803 without requiring @xmath7 to be immersed outside the marked points .
indeed , @xmath1804 is an open subset .
we shall borrow from zehmisch @xcite the notion of _ holomorphic jets _ : given a point @xmath1805 in an almost complex manifold @xmath151 and an integer @xmath1806 , a holomorphic @xmath93-jet at @xmath1805 is an equivalence class of @xmath4-holomorphic curves @xmath1807 with @xmath1808 , where @xmath1809 denotes the @xmath1810-disk in @xmath881 , and two curves are considered equivalent if their partial derivatives at @xmath0 match up to order @xmath93 .
the nonlinear cauchy - riemann equation implies that the holomorphic @xmath93-jet represented by @xmath7 is determined by the holomorphic part of its taylor polynomial of degree @xmath93 ( see @xcite*prop . 2.99 ) , and moreover , every holomorphic taylor polynomial of degree @xmath93 is realizable as the @xmath93-jet of a local @xmath4-holomorphic curve ( @xcite*theorem 2.100 ) .
thus the space of all holomorphic @xmath93-jets at @xmath1805 is a real @xmath1811-dimensional vector space , and the union of these spaces for all @xmath1812 forms a smooth manifold @xmath1813 of real dimension @xmath1814 .
we shall analyze the local structure of @xmath1815 following a minor modification of the scheme outlined in @xcite*chapter 4 . given @xmath1816 representing an element of @xmath1815 , with marked points @xmath1817 , choose a teichmller slice through @xmath551 : this means a smooth @xmath1818-dimensional family @xmath550 of complex structures on @xmath192 that includes @xmath551 and parametrizes a neighborhood of @xmath1819 $ ] in the teichmller space of complex structures modulo diffeomorphisms that are homotopic to the identity and fix @xmath238 . the tangent space @xmath1820 is also required to define a closed complement of the image of the canonical cauchy - riemann operator on @xmath627 restricted to the space of vector fields vanishing at @xmath238 , cf .
@xcite*definition 4.29 .
moreover , we can arrange for @xmath550 to have the following two properties ( cf .
@xcite*lemmas 3.3 and 3.4 ) : * @xmath550 is invariant under the action of the group @xmath1821 of biholomorphic maps on @xmath1822 fixing @xmath238 ; * there exists a neighborhood of @xmath238 on which every @xmath1823 matches @xmath551 .
now let @xmath1824 , and choose any @xmath170 and @xmath171 such that @xmath1825 so the sobolev embedding theorem implies that functions of class @xmath818 on @xmath192 are also in @xmath1826 .
we define the banach manifold @xmath1827 and smooth banach space bundle @xmath1828 with fibers @xmath1829 so that @xmath1830 defines a smooth section .
we say that @xmath1816 is fredholm regular if the linearization @xmath1831 of this section at @xmath1832 is surjective , in which case a neighborhood of @xmath1832 in @xmath1833 is a smooth finite - dimensional manifold , and its quotient by the natural action of @xmath1821 can be identified naturally with a neighborhood of @xmath1834 $ ] in @xmath111 . to incorporate the critical point condition
, fix holomorphic coordinates identifying a neighborhood of each marked point @xmath1835 with the standard unit disk @xmath1836 ; note that this can be done for all @xmath1823 at once since they are assumed to match @xmath551 near @xmath238 . then since @xmath1837 has a continuous inclusion into @xmath1838 ,
there is a well - defined and smooth is clear because it is the restriction to @xmath1833 of a map @xmath1839 which in the natural banach manifold charts provided by @xcite looks like a linear map evaluating derivatives of functions at the fixed points @xmath113 .
this works because we are choosing to represent elements of @xmath111 by maps with marked points at fixed positions ; of course there is no actual constraint on the movement of the marked points , but this freedom is seen in our setup by varying @xmath99 in @xmath550 instead of varying the points @xmath1840 .
this is a notable difference from the setup in @xcite . ]
_ jet evaluation map _ @xmath1841 whose @xmath265th factor for @xmath323 is the holomorphic @xmath322-jet represented by @xmath7 in its parametrization by @xmath1836 at @xmath1842 .
we will say that @xmath1816 is regular for the constrained moduli space @xmath1815 if it is fredholm regular and the jet evaluation map is transverse to the submanifold @xmath1843 consisting of @xmath328-tuples of jets of constant maps .
note that this condition does not depend on the chosen holomorphic coordinates near the marked points , as it is equivalent to the condition that @xmath7 should have vanishing derivatives up to order @xmath322 at @xmath1842 for each @xmath323 .
whenever the regularity condition is satisfied , @xmath1844 inherits the structure of a smooth submanifold with real codimension @xmath1845 , so @xmath1815 in general becomes an orbifold near @xmath1834 $ ] , with @xmath1846 to prove that the constrained regularity condition can be achieved generically , fix @xmath1577 and a suitable sequence of positive numbers @xmath1574 , and consider a banach manifold @xmath556 of almost complex structures in @xmath37 that are @xmath1573-close to @xmath1579 ( cf .
[ sec : sardsmale ] ) .
this gives rise to two universal moduli spaces , @xmath1847 and @xmath1848 where we abbreviate by @xmath1849 the subspaces defined via the condition that @xmath7 be somewhere injective and pass through @xmath35 .
as is well known , @xmath1675 is a separable and metrizable smooth banach manifold , and for @xmath1834 \in { { \mathcal m}}^*_{g , m}(a , j_0)$ ] , a neighborhood of @xmath1680 in @xmath1675 can be identified with the zero - set of a smooth section @xmath1850 where @xmath1851 now denotes the banach space bundle with fibers @xmath1852 the tangent space @xmath1853 is the kernel of the surjective operator @xmath1854 where @xmath1855 is the linearized cauchy - riemann operator associated to @xmath1856 .
we can again define the smooth jet evaluation map @xmath1857 and identify a neighborhood of @xmath1680 in @xmath1858 with @xmath1859 .
the main technical ingredient behind proposition [ prop : critical ] is now the following .
[ lemma : submersion ] the jet evaluation map is a submersion . we need to show that for any @xmath1860 , there exists an element @xmath1861 with @xmath1862 let us first observe that this problem can be solved locally near the marked points : in fact , there exists a smooth section @xmath1863 with @xmath1864 this follows from the local existence theorem for @xmath4-holomorphic curves with prescribed holomorphic derivatives at a point , cf .
@xcite*theorem 2.100 .
more precisely , choose a smooth path @xmath1865 with @xmath1866 and @xmath1867 .
then the local existence theorem provides for each @xmath323 a smooth family of @xmath4-holomorphic curves @xmath1868 defined on sufficiently small disks @xmath1869 such that the holomorphic @xmath322-jet represented by @xmath1870 is @xmath1871 for each @xmath402 .
the desired section @xmath1863 can now be constructed by writing it in our chosen holomorphic coordinates near each marked point @xmath1842 as @xmath1872 and then extending it arbitrarily outside these neighborhoods .
given @xmath662 as above , we aim now to find a pair @xmath1873 such that @xmath1874 in which case @xmath1875 and @xmath1876 .
we will use the weighted sobolev spaces described in [ sec : crpunctured ] .
let @xmath1877 , and assume without loss of generality that @xmath1878 is disjoint from @xmath238 ; this can be achieved at the cost of shrinking @xmath35 and therefore the space of perturbations @xmath556 . as a consequence
, @xmath1879 now has compact support in @xmath218 for any @xmath1710 .
using the fixed holomorphic coordinates on neighborhoods of marked points @xmath1880 , we can identify them biholomorphically with half - cylinders @xmath439 and fix trivializations of @xmath1881 on these neighborhoods to define weighted sobolev norms and a bounded linear map @xmath1882 where sections @xmath662 of class @xmath1883 are required to satisfy @xmath1884 when expressed in the chosen trivialization and holomorphic coordinates @xmath434 on each cylindrical end near @xmath238 . as explained in
[ sec : crpunctured ] , @xmath1885 is asymptotic to the trivial asymptotic operator at each puncture and is thus fredholm for any @xmath1886 .
we claim that whenever this condition is satisfied , the linear map @xmath1887 is surjective .
the proof is more or less standard : we start with the case @xmath765 and note that since @xmath1885 is fredholm , @xmath1888 has closed range , so it is not surjective if and only if there exists a nontrivial section @xmath1889 for @xmath739 which is @xmath628-orthogonal to the images of both @xmath1890 and @xmath1891 .
since @xmath371 has an injective point @xmath1892 with @xmath1893 , the latter implies that @xmath1363 vanishes near @xmath1894 , while the former implies that it is a weak solution to the formal adjoint equation @xmath1895 and is therefore smooth with isolated zeroes , giving a contradiction . the case of general @xmath170 follows from this via elliptic regularity , namely lemma [ lemma : regularity ] .
with this claim in place , we observe that @xmath1896 vanishes near @xmath238 and thus restricts to @xmath218 as a section of class @xmath1897 for any @xmath724 , thus we can find @xmath1898 and @xmath1899 such that @xmath1900 since @xmath1620 has compact support in @xmath218 and @xmath1901 near @xmath238 , this equation implies @xmath1902 near @xmath238 .
the continuous inclusion @xmath1903 implies that @xmath1465 also has a continuous extension over @xmath192 that vanishes on @xmath238 ; moreover , since implies a continuous inclusion @xmath1904 , @xmath1465 has a bounded first derivative on the cylindrical ends , implying via a short computation that for @xmath1905 , the @xmath1906-norm of its derivative on punctured disk - like neighborhoods of @xmath238 is finite .
it follows that the extension of @xmath1465 over the punctures is in @xmath1907 on @xmath192 , and elliptic regularity then implies that it is smooth everywhere .
finally , the exponential weight condition implies that in each holomorphic coordinate system identifying the neighborhood of a marked point @xmath1880 with @xmath1908 such that @xmath1842 is at the origin , we have @xmath1909 for some constant @xmath1910 .
but the choice of @xmath724 in this discussion was arbitrary , so choosing it large enough , we can arrange for @xmath1465 to have vanishing derivatives of arbitrarily large finite order at @xmath238 , proving @xmath1911 .
the lemma implies that @xmath1858 is a separable and metrizable smooth banach manifold , so we can now apply the sard - smale theorem to the projection @xmath1912 giving a baire subset of @xmath556 for which @xmath1913 is a manifold of the correct dimension , and the countable intersection of these subsets for all @xmath69 , @xmath328 , @xmath700 and @xmath1790 is again comeager in @xmath556 , proving that there is a @xmath36-dense subset of @xmath37 for which the statement of the theorem holds . to turn this into a baire subset of @xmath37
, one can use the standard taubes trick ( see e.g. @xcite*4.4.2 ) : present @xmath1913 as a countable union of compact subsets , and associate to each one a set of regular almost complex structures , which is open by construction and dense due to the argument above , so its intersection is comeager .
lemma [ lemma : submersion ] implies that for generic @xmath4 , the jet evaluation map can be made transverse to any given submanifold , hence this method can be used to understand any moduli space of holomorphic curves with marked points satisfying conditions on their derivatives , e.g. the incidence conditions studied by cieliebak and mohnke @xcite .
a statement nearly but not completely identical to proposition [ prop : isolated ] has been proved before by zinger , see @xcite*prop . 3.2 . for the sake of completeness
, we present a self - contained proof in this appendix which is essentially the same as zinger s , but will be easier to follow for readers unaccustomed to the notation in his paper . note that our statement belongs to the almost complex category and makes no reference to any symplectic structure ; the proof however will introduce an auxiliary symplectic structure and use gromov s compactness theorem .
we recall from [ sec : defn ] that if @xmath135 and @xmath139 and @xmath138 are integers , we denote by @xmath1914 the moduli space of all stable nodal @xmath44-fold covers of @xmath7 with arithmetic genus @xmath74 .
suppose @xmath56 is a @xmath36-convergent sequence of almost complex structures on a manifold @xmath32 , and @xmath1915 \in { { \mathcal m}}_g(a , j_\infty)$ ] is a super - rigid curve .
then @xmath1916 is fredholm regular with index @xmath0 , so the implicit function theorem implies the existence of curves @xmath1917 for sufficiently large @xmath59 such that @xmath62 and @xmath63 in @xmath36 ; these curves are unique up to biholomorphic reparametrization , and are also simple and immersed for sufficiently large @xmath59 .
assume @xmath1918 is a sequence of @xmath60-holomorphic curves converging to a nodal cover @xmath1919 for some @xmath1920 .
we will show that if the curves @xmath64 are not covers of @xmath66 for all sufficiently large @xmath59 , then rescaling the normal fibers near @xmath66 as @xmath1921 gives rise to a nontrivial section in the kernel of the normal cauchy - riemann operator on some cover of @xmath1916 , contradicting super - rigidity . choose a convergent sequence of @xmath60-invariant riemannian metrics and corresponding levi - civita connections @xmath1922 .
since the maps @xmath66 are immersed , we can define @xmath60-invariant normal bundles @xmath1923 as the orthogonal complements of @xmath1924 .
these are all isomorphic as real vector bundles , so we can identify them all with the real bundle @xmath1925 carrying a sequence of complex structures @xmath1926 and then use the sequence of exponential maps determined by @xmath1922 to define a @xmath36-convergent sequence of immersions @xmath1927 of some fixed neighborhood @xmath1928 of the zero section @xmath1929 onto some neighborhood of @xmath1930 , such that @xmath1931 . let @xmath1932 for @xmath1933 , so that for @xmath59 sufficiently large , the curves @xmath64 can be identified with @xmath1934-holomorphic curves in the total space of @xmath333 , and each @xmath66 is identified with the zero section .
let @xmath1935 denote the normal projection , so that @xmath1936 induces a connection on @xmath1937 ( as a _ real _ vector bundle ) , and thus defines a splitting into horizontal and vertical subbundles @xmath1938 this splitting is invariant under the diffeomorphisms on @xmath333 defined by real scalar multiplication . for
@xmath1415 and @xmath1939 , the fibers in the splitting admit canonical identifications @xmath1940 and we can write @xmath1934 with respect to the splitting as @xmath1941 for some smoothly varying linear maps @xmath1942 , @xmath1943 and so forth .
since @xmath1944 is @xmath60-holomorphic and the fibers of @xmath1945 are @xmath60-invariant along @xmath66 , we have @xmath1946 now for any constant @xmath1806 , the diffeomorphism @xmath1947 transforms @xmath1934 to @xmath1948 so given any positive sequence @xmath1949 , the sequence @xmath1950 converges in @xmath36 on compact subsets of @xmath333 to @xmath1951 we use a variation on thurston s method for constructing symplectic forms on fibrations ( cf .
@xcite*theorem 6.3 ) . for any open subset @xmath1953 ,
let @xmath1954 denote the space of smooth @xmath53-forms @xmath1363 on @xmath1955 satisfying the following conditions : we observe that @xmath1954 is nonempty whenever there exists a complex trivialization of @xmath1960 over @xmath35 , and moreover , it is _
@xmath36-convex _ in the sense that if @xmath1961 , then @xmath1962 for every smooth function @xmath1963 $ ] .
it follows that an element of @xmath1964 can be constructed by patching together local constructions via a partition of unity .
now given @xmath1965 , choose an area form @xmath1136 on @xmath192 taming @xmath1966 . then for a sufficiently large constant @xmath1967 , @xmath1968 is a closed @xmath195-form that tames @xmath1952 at @xmath192 and hence also in a neighborhood of @xmath192 .
the above proof did not use any special properties of @xmath1952 except that the zero section is pseudoholomorphic and the normal fibers along the zero section are also complex .
the same argument shows that for any embedded closed @xmath4-holomorphic curve in any almost complex manifold @xmath151 , a neighborhood of the curve admits a symplectic form that tames @xmath4 .
[ lemma : normalsections ] suppose @xmath1969 is a smooth map , @xmath100 is a complex structure on @xmath1970 , and @xmath1971 is a smooth section along @xmath183 .
then the resulting map into the total space @xmath1972 is a pseudoholomorphic map @xmath1973 if and only if @xmath1974 is holomorphic and @xmath1975 .
write @xmath1976 . then using
, the equation @xmath1977 translates into the two equations @xmath1978 and @xmath1979 d\psi(z ) \circ \tilde{\jmath } = 0\ ] ] for @xmath1980 .
the first equation says that @xmath1981 is holomorphic , and under this assumption , the second matches @xmath1982 after observing @xmath1983 \circ d\psi \circ \tilde{\jmath } = \pi_n \circ ( \nabla_\eta j_\infty ) \circ
t(u_\infty \circ \psi ) \circ \tilde{\jmath}.\ ] ] we now prove proposition [ prop : isolated ] as follows .
arguing by contradiction , assume after taking a subsequence that the curves @xmath1984 are not covers of @xmath66 for any @xmath59 as @xmath1921 .
choose a symplectic form @xmath30 near the zero section in @xmath1985 as given by lemma [ lemma : tame ] , and choose @xmath724 such that @xmath30 tames @xmath1952 on @xmath1986 .
writing @xmath1987 for sequences @xmath1988 and @xmath1989 , we have @xmath1990 and @xmath1949 by assumption . then @xmath1991 is a sequence of smooth pseudoholomorphic curves in a compact subset of the neighborhood @xmath1986 , which can be written as @xmath1992 where @xmath1993 satisfies @xmath1994 note that since @xmath64 converges to a nodal curve in @xmath1995 , we can also assume the maps @xmath1988 have fixed degree @xmath44 .
then since @xmath1996 and the latter is tamed by @xmath30 in the region under consideration , gromov compactness applies to @xmath1997 and yields a subsequence convergent to a stable nodal curve @xmath1998,\hat{j}_\infty^0)$ ] . by lemma [ lemma : normalsections ] , each smooth component @xmath453 of @xmath1999 has the form @xmath2000 where @xmath1974 is holomorphic and @xmath1982 .
we claim there must be at least one such component for which @xmath2001 and @xmath2002 .
indeed , implies that there is at least one component with @xmath2002 .
if every such component also satisfies @xmath2003 , then @xmath662 is a nonzero constant on this component , as the normal operator @xmath2004 is simply the standard cauchy - riemann operator on a trivial bundle when @xmath183 is constant .
but since @xmath2005 , any component with @xmath2003 is necessarily connected by a chain of nodes to another component with @xmath2001 , and on this component , @xmath662 is nonzero at the nodal point .
this implies the existence of a nontrivial element @xmath2006 for some positive degree holomorphic cover @xmath183 , and thus violates super - rigidity .
the proof of proposition [ prop : isolated ] is complete . | we develop new techniques to study regularity questions for moduli spaces of pseudoholomorphic curves that are multiply covered . among the main results ,
we show that unbranched multiple covers of closed holomorphic curves are generically regular , and simple index @xmath0 curves in dimensions greater than four are generically super - rigid , meaning their branched covers form connected components of the moduli space .
we also establish partial results on super - rigidity in dimension four and regularity of branched covers , and briefly discuss the outlook for bifurcation analysis . the proofs
are based on a representation - theoretic splitting of cauchy - riemann operators with symmetries . |
the sloan digital sky survey ( sdss ; york et al .
2000 ) is using a drift - scanning imaging camera and two multi - fiber double spectrographs on a dedicated 2.5 m telescope ( gunn et al .
1998 ) to image 10,000 of sky on the sdss @xmath4 ab magnitude system ( fukugita et al . 1996 ; stoughton et al . 2002 ) .
spectra will be obtained for @xmath010@xmath5 galaxies to @xmath6 and @xmath010@xmath7 quasars to @xmath8 ( @xmath9 for @xmath10 candidates ) .
quasar candidates are targeted for spectroscopy based on color criteria ( richards et al .
2002 ) or because they are unresolved objects with radio emission detected by the first survey ( becker , white , & helfand 1995 ) .
the essence of the color selection is simple : target objects whose broad - band colors are different from those of normal stars and galaxies , especially objects with colors similar to those expected for known and simulated quasars ( richards et al .
objects in the few untargeted outlying regions of color space , plus first sources with either resolved or unresolved counterparts , can also be selected to @xmath11 as _ serendipity _ targets ( stoughton et al .
2002 ) . due to these inclusive criteria ,
the selection of candidates using @xmath12 band magnitudes rather than blue magnitudes which are more affected by absorption and reddening , and its area and depth , the sdss is proving effective at finding unusual quasars .
many of these unusual quasars are unusual broad absorption line ( bal ) quasars .
bal quasars show absorption from gas with blueshifted outflow velocities of typically 0.1@xmath13 ( weymann et al .
about 10% of quasars exhibit bal troughs , but this is usually attributed to an orientation effect .
most quasars probably have bal outflows covering @xmath010% of the sky as seen from the quasar , with mass loss rates possibly comparable to the accretion rates required to power the quasar ( @xmath14 @xmath15yr@xmath16 ) .
therefore an understanding of bal outflows is required for an understanding of quasars as a whole .
unusual bal quasars may help in this endeavor since they delineate the full range of parameter space spanned by bal outflows .
examples of unusual bal quasars with extremely strong or complex absorption whose nature is difficult to discern at first glance have been found in the past few years through followup of first radio sources ( becker et al .
1997 , 2000 ) and @xmath17 quasar candidates from the digitized palomar sky survey ( djorgovski et al .
2001 ) . in this paper
we show that the sdss has confirmed that these are not unique objects , but members of populations of unusual bal quasars .
these populations include extremely reddened objects and objects with unprecedentedly strong absorption , absorption from record numbers of species , absorption from unusual species or with unusual line ratios , or absorption which extends longward of the systemic redshift .
we begin in [ bkgd ] with an overview of our current knowledge of bal quasars , to put in context the implications of these unusual objects .
we discuss our sample of sdss quasars in [ data ] and our selection of a sample of unusual bal quasars from it in [ select ] . in [ analysis ] we present the different categories of unusual bal quasars we have identified . in [ discussion ]
we discuss the implications of these objects for models of bal outflows .
we summarize our conclusions in [ conclusions ] . in appendix [ bi ]
we discuss how we measure the strength of bal troughs for sdss quasars .
to understand why our bal quasars are unusual and important , we briefly review the properties of ` ordinary ' bal quasars and some key current questions in bal quasar research . for more extensive discussions , see ( 1997 ) and
other contributions to , shlosman , & weymann ( 1997 ) , ( 2000 ) , and the many contributions to , kraemer , & george ( 2001b ) .
observationally , bal quasars show troughs @xmath02,000@xmath1820,000kms@xmath16 wide arising from resonance line absorption in gas with blueshifted ( outflowing ) velocities up to 66,000 kms@xmath16 ( foltz et al .
the absorption troughs are _ detached _ @xmath020% of the time ( turnshek 1988 ) , meaning that the onset redshift of the absorption lies shortward of the systemic redshift , by up to 50,000 kms@xmath16 ( jannuzi et al . 1996 ) .
note that we use positive velocities to indicate blueshifts since bal troughs are outflows .
the usual formal definition of a bal quasar is a quasar with positive _
balnicity index_. the bi is a measure of the equivalent width of the civ absorption defined in the seminal paper of ( 1991 ) , hereafter w91 .
however , this criterion is not perfect ; for example , it ignores absorption @xmath192000kms@xmath16 wide even though many such _ mini - bals _ are now known to share most or all of the other characteristics of bal quasars ( hamann 2000 ) .
we discuss this issue further in appendix [ bi ] , where we define the _ intrinsic absorption index_. the ai is a refined bi designed to make optimal use of sdss spectra and to include as bal quasars objects which were previously excluded but are clearly related to traditional bal quasars .
bal quasars are divided into three observational subtypes depending on what type of transitions are seen in absorption .
a list of relevant transitions with rest frame @xmath20@xmath211215 is given in table [ t_lines ] , and two of these three subtypes are illustrated and discussed in figure[f_examples ] .
high - ionization bal quasars ( _ hibals _ ) show absorption from civ , nv , siiv , and ly@xmath22 ( in order of decreasing typical strength , for @xmath20@xmath211215 ) .
low - ionization bal quasars ( _ lobals _ ; , weymann , & korista 1993 ) exhibit the above high - ionization absorption plus absorption in mgii , aliii , alii and sometimes feii , feiii and other low - ionization species , again roughly in order of decreasing strength . note that the absorbing gas has a range of ionization states even in hibals .
the observed range simply extends to lower ionization states in lobals .
the rare iron lobals ( _ felobals _ ; 1987 ; cowie et al . 1994 ; becker et al .
1997 , 2000 ; menou et al .
2001 ) also show absorption from excited fine - structure levels or excited atomic terms of feii or feiii . , @xmath23 , and @xmath24 , and
consists of ( 2@xmath24 + 1)@xmath25(2@xmath23 + 1 ) energy _ states _ grouped into _ levels _ of different total angular momenta @xmath26=@xmath23+@xmath24 . though it is a misnomer , we follow convention and use _ excited - state _ to refer to absorption from excited fine - structure levels , excited terms , or both .
note that absorption arising from excited fine - structure levels of the ground term of feii is sometimes denoted feii * , though technically feii * to feii * * * * should be used , depending on the excitation ( morton 1975 ) , and similarly for other ions .
] a key feature of bal outflows which has only recently been appreciated is that the absorption is saturated ( optical depth @xmath27 of a few ) in almost all cases where optical depths can be determined using line ratios and other saturation diagnostics ( e.g. , arav et al .
this is true even though the absorption troughs rarely reach zero flux .
this ` nonblack saturation ' means that the bal outflows typically exhibit _ partial covering _ of the continuum source ( arav et al .
1999a ) , possibly combined with infill of the troughs by host galaxy light or scattered light which bypasses the bal outflow ( ogle et al . 1999 ) . the bal outflow is usually thought to lie outside the broad emission line region since both the continuum and the broad emission lines are absorbed in most objects ( turnshek et al .
however , in some cases , only the continuum emission is absorbed ( arav et al .
1999a ) .
species of a given ionization tend to have similar but not identical trough shapes in velocity space , due to similar but not identical partial coverings as a function of velocity for different ions ( e.g. , arav et al .
when present , low - ionization troughs are always narrower than high - ionization troughs , are usually strongest at low outflow velocities ( 1993 ; but see 2001a ) , and are often seen in relatively narrow features where the high - ionization absorption is strongest .
this latter feature indicates that the low and high ionization regions are closely associated ( e.g. , de kool et al .
large local density gradients are probably required for gas with ionization parameter @xmath28 ranging at least from 10@xmath29 to 1 ( arav et al .
2001b ) to exist in such close proximity ( arav et al .
bal troughs have never been seen to vary in velocity at a significant level .
limits on the acceleration reach @xmath190.03cms@xmath29 in one case ( wampler , chugai , & petitjean 1995 ) , and the most significant variations claimed are only @xmath30kms@xmath16 over 5 rest - frame years ( @xmath31=0.035cms@xmath29 ; & irwin 2001 ) and @xmath32kms@xmath16 over 6.6 rest - frame years ( @xmath31=0.06cms@xmath29 ; , veilleux , & sanders 2002 ) . thus the bal gas must either be coasting or in a stable flow pattern with a significant transverse velocity component ( arav et al .
bal troughs have been seen to vary in strength ( [ var ] ) , probably due mostly to variations in covering factor rather than ionization , since bal troughs are typically saturated with optical depths @xmath27 of a few ( e.g. , 2001b ) .
bal quasars form 8% of the optically selected flux - limited large bright quasar survey , but 11% after accounting for the average differential optical / uv @xmath33correction between bal and non - bal quasars caused by the absorption troughs ( weymann 1997 ) .
lobals comprise @xmath015% of bal quasars ( w91 ) , but the differential @xmath34-correction between lobals and hibals could be substantial ( sprayberry & foltz 1992 ) . if all quasars can be bal quasars when viewed along certain lines of sight , then the @xmath34-corrected bal quasar fraction should be equal to the average global covering factor ( gcf ) of those lines of sight ( morris 1988 ) .
the @xmath34-corrections used to estimate the true bal quasar fraction must also account for any differential attenuation between bal and non - bal quasars .
it has been widely held that hibals as a population are not heavily reddened ( e.g. , w91 ) but that lobals are ( sprayberry & foltz 1992 ) , and thus that the true fraction of lobals is underestimated by flux - limited optical surveys .
other surveys can overcome this bias : @xmath35% of _ all _ iras - selected quasars are lobals ( w91 ; low et al .
1989 ; boroson & meyers 1992 ) and @xmath36% of bal quasars are lobals in the optical and radio flux - limited first bright quasar survey ( fbqs ) , which has an observed bal quasar fraction of 14@xmath374% ( 18@xmath374% if the balnicity criterion is relaxed ; becker et al .
the fbqs and the sdss also find that hibals as well as lobals are reddened ( brotherton et al .
2001b ; menou et al .
thus all types of bal quasars are underrepresented in optical flux - limited samples and the true average gcf must be @xmath210.11 .
a gcf of @xmath00.3 would help explain why bal quasars are typically more polarized than non - bal quasars ( schmidt & hines 1999 ; hutsem ' ekers & lamy 2000 ) and have generally weaker narrow [ oiii ] emission ( boroson & meyers 1992 ) .
an upper limit to the gcf can be estimated since broad absorption line troughs are really broad scattering troughs : absorption of these resonance line photons is followed by emission of an identical photon .
if the gcf was unity , then the ` absorbed ' flux would be redistributed to a red wing in each line , and the emission and absorption equivalent widths ( ew ) would be identical ( hamann , korista , & morris 1993 ) . since the absorption ew is typically much greater than the emission ew , the bal gas must have [email protected] .
however , this inference ignores the very real possibility of preferential destruction of line photons by dust , particularly in lobals ( voit et al .
at least some lobals meet every test for being recently ( re)fueled quasars in the act of casting off their cocoons of dust and gas " ( voit et al .
1993 ) . such dust , combined with near - unity gcfs , helps explain many of their unusual properties ( canalizo & stockton 2001 ) .
the scattering of resonance line photons can provide the radiative acceleration which at least partially drives bal outflows ( arav et al .
the difficulty arises in accelerating the gas to sufficient velocities without completely stripping the resonance - line - absorbing ions of their electrons . the disk wind model of murray & chiang ( 1998 , and
references therein ) has been very successful in explaining this and other properties of bal quasars . in this model
, a wind from an accretion disk is shielded from soft x - rays by a high column density ( @xmath39 ) of highly ionized gas ( @xmath40 ) .
being stripped of many outer electrons , the ions in this ` hitchhiking ' gas transmit uv photons from resonance lines such as civ .
thus the wind can be accelerated by scattering such photons without becoming too highly ionized by higher energy photons .
the ` hitchhiking ' gas successfully explains the strong soft x - ray absorption in bal quasars ( @xmath41cm@xmath29 ; e.g. , 2001a ) and absorption from very high ionization lines ( up to neviii , mgx , and sixii in sbs 1542 + 541 ; 1998 ) .
the quasar structure proposed by ( 2000 ) also posits a disk wind , but from a narrow range of radii , such that bal quasars are only observed when the line of sight is directly aligned with the wind .
a recent summary of theoretical and computational modeling of disk winds can be found in , stone , & kallman ( 2000 ) .
some bal quasars , particularly lobals , may be quasars cocooned by dust and gas rather than quasars with disk winds ( becker et al .
2000 ) , but the only serious modeling relevant to this alternative has been the work of , baker , & perry ( 1999 ) . it is clear that broad absorption line quasars remain an active area of research . disk wind models explain many properties of bal quasars , but it is unclear if they can explain the full range of bal trough profiles and column densities .
the average global covering factor of bal gas and the range of global covering factors remain uncertain .
thus it is an open question how many bal quasars are normal quasars seen along special lines of sight , and how many are young quasars emerging from dusty ` cocoons ' .
bal quasars with unusual and extreme properties may be useful in answering some of these questions .
whenever possible , the spectral and photometric data presented here are taken from the sdss early data release ( edr ) , since it was produced with essentially a single uniform version of the sdss data processing pipeline . a detailed description of the edr observations and reductions is given by ( 2002 ) .
here we review only a few points relevant to our primarily spectroscopic analysis .
sdss spectra are obtained using plates holding 640 fibers , each of which subtends 3 on the sky , and cover 38009200 with resolution of @xmath43 and sampling of @xmath442.4 pixels per resolution element .
the relative spectrophometric calibration is good to @xmath010% , but the absolute spectrophometric calibration to only @xmath030% .
many objects have spectra from multiple plates for quality assurance purposes . all spectra presented here are coadditions of all available spectra using inverse variance weighting at each pixel . unless otherwise indicated , all spectra are from the sdss ; the exceptions are from the keck , cfht , and arc 3.5 m telescopes .
all magnitudes in this paper are _ psf magnitudes _ calculated by fitting a model psf to the image of the object and correcting the resulting magnitude to a 74 aperture ( 2001 and 4.4.5 of 2002 ) .
the sdss uses asinh magnitudes ( lupton , gunn , & szalay 1999 ) which differ from traditional magnitudes by @xmath191% for signal - to - noise ratios snr@xmath4510 . because the photometric calibration is still uncertain at the @xmath05% level , all magnitudes are provisional and
are denoted using asterisks .
the astrometric calibration is good to 01 rms per coordinate .
iau designations for each object presented here ( e.g. , sdss j172341.09 + 555340.6 ) are given in the tables , but are shortened in the text for brevity ( e.g. , sdss 1723 + 5553 ) .
full iau designations given in the text refer to quasars in the edr quasar catalog ( schneider et al .
visual inspections were carried out to flag unusual spectra which were not immediately identifiable as normal quasars , hibals , lobals , or any other known type of object or spectral reduction problem .
sfa and/or pbh inspected all 95 spectroscopic plates in the edr , all 43 other plates with numbers less than 416 the highest number in the edr completed on or before modified julian date ( mjd ) 52056 , and a selection of 69 ` post - edr ' plates numbered 417 and up . also , all edr spectra not targeted and spectroscopically confirmed as galaxies were inspected by dev and/or aeb , and all edr objects spectroscopically classified as quasars were re - inspected by pbh .
many other workers also inspected varying quantities of plates so that each spectrum has likely been inspected three times .
most of the unusual objects found were unusual bal quasars . from @xmath0120,000
spectra on 207 spectroscopic plates , including @xmath08,000 quasar spectra , we have identified eighteen unusual bal quasars and two mysterious objects which might be unusual bal quasars ( [ gradual ] ) .
three additional unusual bal quasars were selected from sdss images but were confirmed with spectra obtained with other telescopes .
all these unusual bal quasars can be divided into a handful of categories , as discussed in the next section .
we are confident that there are no further examples of these categories of unusual bal quasars on the inspected plates among spectra which have snr@xmath466 per pixel in at least one of the @xmath47 , @xmath48 or @xmath12 bands .
this is the only sense in which our unusual bal quasar sample could be considered a complete sample .
a quasar sample from all the plates we inspected has not yet been defined , and the edr quasar sample ( schneider et al . 2002 ) is incomplete because not all quasar candidates in the edr area have been observed spectroscopically and because the quasar candidate selection criteria were not the same for all spectroscopic observations in the edr .
nonetheless , since it is of interest to know what fraction of the quasar population unusual bal quasars comprise , in [ percent ] we estimate this fraction for the edr .
table [ t_info ] gives general information on the unusual bal quasars presented in this paper .
in particular , the target code shows whether or not the object was targeted by the quasar , first , rosat , serendipity and star selection algorithms ( see the note to the table , and 2002 ) .
none of our objects were selected as galaxy targets , or as rosat targets .
only three objects were not selected as quasar targets , and one of them ( sdss 1730 + 5850 ) was not selected only because of its faintness .
only one target ( sdss 0127 + 0114 ) was completely overlooked by color selection and was selected only as a first target .
the efficiency of sdss color selection in finding unusual bal quasars is shown in figure[f_ccdiag ] .
the projections of the stellar locus in the three sdss color - color diagrams are shown , along with the unusual bal quasars presented in this paper , coded according to category .
most of the scatter in colors is due to the bal troughs .
most of these objects have colors quite distinct from ordinary quasars as well as ordinary stars .
the sdss is sensitive to these unusual objects because the sdss quasar target selection algorithm selects objects with unusual colors even if they lie far from the locus of ordinary quasars in color space .
all sdss and followup spectra presented herein are available for download from the contributed data section of the sdss archive at http://archive.stsci.edu / sdss/. flux densities @xmath49 in these spectra , and in all our figures , are given in units of @xmath50ergs cm@xmath29s@xmath16@xmath16 .
all of the unusual bal quasars found in sdss to date are lobals , and many are felobals .
most can be placed into one of five categories , which we now present and analyze separately .
first we present three felobals with _ many narrow troughs _ ( [ scalloped ] ) .
objects with similar numbers of troughs but higher terminal velocities could have their continua almost totally absorbed ; we present five of these _ overlapping - trough _ felobals in [ abrupt ] .
after that we present nine heavily reddened lobals ( [ reddened ] ) and one slightly less reddened lobal with stronger feiii than feii absorption ( [ fe3 ] ) , along with several possibly similar objects with lower snr .
lastly , we present two mysterious objects which may be very unusual bal quasars ( [ gradual ] )
. the implications of all these objects are discussed in
[ discussion ] .
absorption from many different ions is present in these objects . to identify most lines
, we used the wavelengths tabulated in ( 1962 ) , , york , & jenkins ( 1988 ) , ( 1991 ) and ( 2001 ) .
absorption and occasionally emission from many multiplets of feii , feiii , and possibly fei is present in many of these objects as well .
( recall that a _ multiplet _ is the set of all transitions between two atomic terms . ) to identify these lines we also used wavelengths tabulated in ( 1950 ) , ( 1981 ) , ( 1987 ) and , clowes , & campusano ( 1996 ) .
these wavelengths are generally accurate enough for use with the low - resolution spectra presented here , but more accurate wavelengths are available in the feii and feiii literature referenced by & wilkes ( 2001 ) .
a short word about multiplet notation is in order .
the standard notation of ( 1950 ) designates feii multiplets either as ultraviolet ( uv ; @xmath51 ) or optical ( opt ; @xmath52 ) .
both lists are ordered separately , first by increasing energy of the lower term and then by increasing energy of the upper term .
thus uv2 multiplet transitions are to a more energetic upper term than uv1 , though they arise from the same lower term .
the higher the multiplet number , the higher the excitation potential ( ep ) of its lower term ( this applies to all ions , not just feii ) .
for example , feii multiplets uv1 to uv9 arise from the ground term .
when identifying which feii multiplets are responsible for any observed absorption , we always begin with the lowest - numbered multiplet present in that wavelength range .
absorption from higher - numbered multiplets ( i.e. , more highly excited terms ) may be present as well ( or instead , if selective pumping is at work ) , but we prefer to be conservative in estimating the feii excitation . where useful , bal troughs are plotted in velocity space using @xmath53 ( foltz et al .
the uncertainty on @xmath1 is given by @xmath54 we define the zero velocity for each line using its laboratory vacuum rest wavelength .
our inspection revealed three felobals with many narrow absorption troughs , similar to first 0840 + 3633 ( becker et al .
we discuss the spectra of the high - redshift and low - redshift objects separately , and then in comparison , but defer discussion of the implications of all these objects to [ impscal ] .
this object ( figure [ f_hiscal ] ) may have the most absorption troughs ever observed at this resolution in a single quasar .
we identify absorption from 15 ions of 10 elements : alii , aliii , cii , civ , crii , fei , feii , feiii , mgii , niii , nv , oi , siii , siiv , and znii .
feii absorption is seen from multiplets up to at least uv79 , whose lower and upper terms are 1.66ev and 7.37ev above ground , respectively .
only a small amount of neutral gas is probably present , since the only neutral absorption we detect is oi@xmath551302 ( intrinsically one of the strongest neutral - atom transitions in the ultraviolet longward of ly@xmath22 ) and weak fei@xmath552484 .
we adopt an emission - line redshift of @xmath56 by fitting gaussians to the apparent narrow emission lines of [email protected] and [email protected] . both lines used to derive the redshift are the longer - wavelength lines of doublets , but the apparent emission is narrower than the doublet separations , allowing a single line to be fitted . the systematic uncertainty on our adopted redshift dwarfs the statistical uncertainty .
the bottom panel of figure [ f_hiscal ] shows the object s spectrum at @xmath57 in the rest frame at the emission - line redshift ; the spectrum at @xmath58 is discussed in
[ compscal ] .
there are at least three different redshift systems in the absorbing region .
the lowest redshift system has @xmath59 as measured by feiiiuv34 , which is much stronger in this system than in either of the other two .
the intermediate redshift system , the weakest , has @xmath60 from niii and siii . the highest redshift system has @xmath61 from niii , siii , znii , and crii ; it also has weak fei@xmath552484 and probably oi@xmath551302 .
absorption from transitions seen in both the lowest and highest redshift systems is indicated by the bifurcated dotted lines in figure [ f_hiscal]b .
the high - ionization siiv and civ troughs show some features corresponding to the low - ionization redshift systems , but overall extend smoothly to outflow velocities of 11300kms@xmath16 .
the narrow troughs in this object enabled the identification of the absorption just longward of civ as feiiuv44,45,46 absorption from terms with [email protected] .
such absorption has been seen before in the original felobal q 0059@xmath182735 ( wampler et al .
1995 ) , but is much stronger here .
figure [ f_scalloped2 ] shows the full sdss spectra of both low redshift bal quasars with many narrow troughs .
[ [ sdss11250029 ] ] sdss 1125 + 0029 + + + + + + + + + + + + + + this object has narrow [ oii]@xmath623727,3729 emission at @xmath63 , in excellent agreement with the peak of the broader [ neiii]@xmath553869 emission .
broad emission is visible in h@xmath64 and h@xmath1 , at @xmath65 ( possibly affected by superimposed absorption ) ; the apparent broad emission at 7600@xmath187640 ( marked @xmath66 in figure [ f_scalloped2]a ) is just poorly removed telluric absorption .
visible in the red half of the spectrum are relatively narrow absorption lines from hei@xmath553889+h8 , caii@xmath553934(k ) , h@xmath67+caii@xmath553969(h ) blended with [ neiii]@xmath553969 emission , h@xmath68 , h@xmath64 , and h@xmath1 .
all of these lines except caiik show reversed profiles : narrow emission in the center of a broader absorption trough .
these narrow balmer emission lines give @xmath69 , which we adopt as the systemic redshift .
this is in excellent agreement with the caiik redshift and agrees with the caiih redshift within the errors due to blending with hei .
thus the [ oii ] and [ neiii ] emission lines are blueshifted by 310@xmath37130kms@xmath16 and the broad balmer lines by 1030@xmath37320kms@xmath16 .
the bal outflow might contribute to the caii absorption ( cf .
[ abrupt ] ) , but if the caii is due to starlight it implies a host galaxy with @xmath70 , which is extremely luminous .
this object has broad emission in h@xmath1 but not in mgii .
this could be due to preferential destruction of mgii line photons by dust , since resonant scattering can greatly increase the path length for line photons ( voit et al .
bal troughs present in this object include mgii and possibly oiii@xmath553133 and hei@xmath553188 .
the vast majority of the remaining absorption at @xmath71 is from feii .
[ [ sdss11280113 ] ] sdss 1128 + 0113 + + + + + + + + + + + + + + this object has @xmath72 as measured by narrow [ oii]@xmath623727,3729 emission and somewhat broader [ neiii]@xmath553869 , [email protected] and [ neiii]@xmath553969 .
h@xmath64 is present and may be slightly redshifted , but the snr is low and it may be blended with [ oiii]@xmath554363 .
h@xmath1 is also present , but we can not determine its peak observed wavelength with any certainty since it is at the extreme red end of our spectrum .
we can not identify the apparent moderately broad emission line just shortward of h@xmath64 , at 4295 ( marked ? in figure [ f_scalloped2]b ) ; there should be other lines present if it is feii emission . the bal troughs in this object
are not quite as omnipresent as those in sdss 1125 + 0029 .
the similarity of our many - narrow - trough bal quasars to each other is shown in figure [ f_scalloped3 ] , where we plot the wavelength range 20003300 in the rest frame of the deepest absorption trough for each object .
dotted vertical lines show the wavelengths of the labelled absorption lines or numbered feii multiplets in this frame .
there is no sign of mgi@xmath552852 absorption in any of the objects .
dashed vertical lines show the wavelengths of mgii@xmath622796,2803 , hei@xmath553188 , and two oiii transitions ( see next paragraph ) in the adopted _ systemic _ rest frame of each object .
note that if they are present , hei@xmath553188 is blended with feiiopt6 and opt7 and the weaker hei@xmath552945 line , is blended with feiiuv60 and uv78 .
no absorption trough reaches zero flux in any object , indicating partial coverage of the continuum source by the absorbing region , and possibly a contribution from scattered light which bypasses the absorbing region .
sdss 1125 + 0029 ( top ) has the lowest partial covering but the highest feii excitation , since it shows absorption which arises only from highly excited feii , seen longward of mgii and between the more common lower excitation feii absorption troughs at 2400 and 2600 ( moore 1950 ) .
most troughs can be identified with feii multiplets , but some have uncertain identifications , notably : + @xmath73 we initially identified the @xmath03100 - 3135 trough with oiii@xmath553133 , which was first identified as a bal in cnoc2 j022509.6 + 001904 ( hall et al .
this transition has a lower state 36.9 ev above ground , but can be indirectly populated via bowen resonance - fluorescence with heii ly@xmath22 ( section 4.7 of 1989 ) .
however , in this case we should see oiii@xmath553445 with @xmath030% the strength of oiii@xmath553133 .
such absorption is not present in either object in figure [ f_scalloped2 ] , so oiii@xmath553133 absorption can at best explain only part of their @xmath03100 - 3135 troughs .
the lowest - numbered feii multiplet which matches this trough is opt82 ( ep 3.87ev ) , but if that identification were correct we should also see absorption near 2150 from multiplet uv213 ( ep 3.14 - 3.22ev ) .
another possibility is crii absorption from terms
@xmath02.45ev above ground ( de kool et al .
2002 ) , but in sdss 1125 + 0029 such absorption would be improbably strong relative to the ground - term crii trough at @xmath02060 .
thus the origin of the @xmath03100 - 3135 trough remains unclear .
+ @xmath73 explaining the @xmath02220 trough as feii requires absorption from at least uv168 ( lower term ep [email protected] ) .
the 2400@xmath182550 absorption can then be attributed to uv144@xmath18149,158@xmath18164 .
such absorption is weak in sdss 1128 + 0113 and absent in sdss 1723 + 5553 , however , so their 2220 troughs are unlikely to be due solely to feiiuv168 .
a contribution from niii ( table [ t_lines ] ) is a poor fit , and sii and mni can be ruled out because many other lines from neutral atoms would also be seen and are not .
+ @xmath73 the strong absorption near @xmath02665 may require more than just criiuv7,8 to explain it .
+ the first two cases above might be explained by selective pumping of the lower terms of multiplets opt82 and uv168 , respectively , a possibility which can only be tested via detailed modeling of feii .
the most interesting feature of both low - redshift objects presented in [ loscal ] is that the broad mgii doublet absorption trough extends longward of the systemic redshift ( dashed vertical lines in figure[f_scalloped3 ] ) . in sdss 1128 + 0113 , the mgii absorption appears split into two components , but this structure may be some combination of narrow mgii emission and partial covering which varies with velocity .
the presence of longward - of - systemic broad absorption is quite surprising , since the absorption in bal quasars has always previously been seen in outflow .
a search for bal troughs longward of the systemic redshift in all bal quasars in the edr ( [ percent ] ) turned up several additional candidates , but none stood up to scrutiny .
sdss [email protected] has @xmath74 from narrow [ oii]@xmath623727,3729 emission .
mgii absorption extending slightly longward of this @xmath75 can not be ruled out due to noise from the 5577 night sky line , but there is no evidence of longward - of - systemic absorption from other lines .
sdss [email protected] has @xmath76 from narrow [ oii ] emission .
there is weak mgii emission in the middle of what appears to be an absorption trough but is actually just a gap between emission from feii multiplets .
sdss j235238.09
+ 010552.4 has civ absorption longward of the peak of civ emission , but only because the civ peak is blueshifted @xmath04000kms@xmath16 from the ciii ] and mgii redshift of @xmath77 .
we discuss the implications of longward - of - systemic absorption in [ impscal ] .
several felobals have been discovered from the sdss with abrupt drops in flux near mgii@xmath622796,2803 caused by overlapping absorption troughs .
that is , the troughs remain deep at velocities comparable to the spacing of absorption troughs from different transitions , so that there are no continuum windows between the troughs ( figures [ f_0300]@xmath18[f_weird2 ] ) .
troughs 19000kms@xmath16 wide may be required , so that mgii overlaps feiiuv1 at @xmath55@xmath782632 ; however , a width of only 13000kms@xmath16 is needed if feiiuv62,63 absorption is present .
these ` overlapping - trough ' bal quasars resemble mrk 231 ( smith et al . 1995 ) , first 1556 + 3517 ( becker et al .
1997 ) and especially fbqs 1408 + 3054 ( white et al . 2000 ; becker et al . 2000 ) , including the complex emission and absorption near mgii and the strong mgi@xmath552852 absorption .
however , the bal region in our overlapping - trough objects covers the emission region almost totally , instead of only partially as in fbqs 1408 + 3054 or mrk 231 .
( the spectrum of fbqs 1408 + 3054 can be seen in [ gradpc ] ) .
we first discuss the rest frame features common to all these objects ( figures [ f_0300]@xmath18[f_weird2 ] ) , and then present them individually , in order of increasing redshift .
the apparent emission lines shortward of mgii are identified as regions near the high - velocity ( blue ) ends of various absorption troughs . at high outflow velocities , the partial covering of the absorption , or possibly its optical depth ,
typically decreases so that the observed flux begins to recover to the continuum level .
the troughs are broad enough , however , that before this recovery is complete , another absorption trough is encountered and the observed flux drops abruptly .
these abortive recoveries toward the continuum level can mimic the appearance of broad emission lines .
longward of mgii , the spectra are mostly free from absorption and we can identify emission features by comparison to the sdss composite quasar spectrum ( vanden berk et al . 2001 ) , allowing for emission from strong fei and feii multiplets whose lower terms are @xmath383ev above ground . in our object with the best coverage longward of mgii ( figure [ f_0300 ] )
, we see h@xmath1 in emission , feiiopt37,38 at 4570 , h@xmath64+[oiii ] at 4350 , feiopt5,21 at 3735 , feiopt23 emission at 3600 , a feature at 3350@xmath183550 including feiopt6,81 at 3465,3435 and possibly feiiopt4,5 at 3500,3400 , a feature at 3150@xmath183300 including feiopt91+feiiopt1 at 3280 and hei+feiiopt6,7+feiopt155 - 158 at 3200 , and a blend of feiopt9,30 at 3015 with feiiuv60,78+feiuv1 at 2970 . many of these features are also visible in figures [ f_1154]-[f_weird2 ] . near mgii
the spectra are blends of emission and absorption
. there may be hei , fei and/or feii absorption shortward of the 3200 emission feature in some or all of the objects , most notably in sdss 1730 + 5850 ( figure [ f_abruptbals]b ) .
there is almost certainly fei and/or feii absorption shortward of the 2970 emission feature , because the observed flux dips lower than essentially any point in the continuum longward of mgii .
dust may contribute to this drop in flux , but the effect is too sudden to be entirely due to dust .
also , in all objects except sdss 0819 + 4209 ( figure [ f_abruptbals]a ) , the flux recovers to this level at least once shortward of mgii , which would not be the case if dust reddening were already affecting the spectrum strongly at 3000 .
mgii@xmath622796,2803 emission must also contribute to the spectrum in this region , but mgi@xmath552852 and mgii absorption are so strong and abrupt that only a narrow sliver of probable mgii emission remains , just longward of the onset of those troughs .
shortward of mgii , the overlapping troughs make detailed line identifications quite difficult .
however , we know that these objects are felobals because they show feiiuv62,63 absorption at 2750 ( e.g. , figure [ f_0300 ] ) and because they show feii*uv1 absorption .
the latter can be identified because absorption from excited levels in the uv1 multiplet extends to 2632 , while ground level uv1 absorption extends only to 2600 .
given low - ionization troughs 13,000@xmath1819,000kms@xmath16 wide or wider , feii absorption can blanket the spectrum down to at least 2100 .
a recovery of flux is detectable in all these objects around the ciii]@xmath551908 line , from 2100 down to the onset of aliii absorption at 1860 .
aliii blended with feii and other low - ionization lines ( cf .
figure [ f_hiscal ] ) can overlap with alii and still more feii all the way down to civ .
then , given that high ionization troughs are typically broader than low ionization troughs , it is not surprising that the civ trough overlaps with siiv and siiv with nv , so that the entire spectrum down to ly@xmath22 is essentially extinguished .
we assume the continuum in these objects shortward of mgii is flat in @xmath49 at the level of the 3100 window ( indicated by a short dot - dashed line in each figure ) .
this window may be affected by absorption from hei@xmath553188 , feii or even oiii@xmath553133 , as well as by reddening , but in the three objects with significant coverage longward of it , this window is not a bad match to the continuum .
we use 19000kms@xmath16 as the upper velocity limit when calculating the ai and bi from mgii in most of these objects ( appendix [ bi ] ) .
the maximum ai in that case is 19000kms@xmath16 , while the maximum bi depends on the detachment velocity of the bal trough . for sdss 0437@xmath180045
, we use civ to measure the ai and bi ( maximum values 25000kms@xmath16 and 20000kms@xmath16 respectively ) since our spectra fully cover the civ region in that object .
in addition to an extremely broad mgii bal trough , sdss 0300 + 0048 ( figure [ f_0300 ] ) shows a strong caii h&k bal trough . this caii absorption appears split into two relatively narrow systems , at @xmath79 and @xmath80 ( @xmath81=1550kms@xmath16 ) , although there is also broad caii absorption extending a further 2000kms@xmath16 shortward .
there is associated mgii and mgi@xmath552852 absorption 2300kms@xmath16 longward of the highest redshift caii system , at @xmath82 .
we have adopted this latter value as the systemic redshift . associated narrow mgii systems with @xmath83
do exist , but 2300kms@xmath16 would be an extreme velocity for such a system ( foltz et al .
1986 ) , whereas bal troughs detached by 2300kms@xmath16 shortward of @xmath84 are unremarkable .
caiih&k absorption in bal outflows has been seen before only in the seyfert1/lobal mrk 231 ( boksenberg et al .
1977 ) and the felobals fbqs 1044 + 3656 ( white et al .
2000 ) and q 2359@xmath181241 ( arav et al .
sdss 0300 + 0048 is also a felobal , with feii absorption at 2750 and near 2400 and 2600 .
however , the feii bal trough is associated only with the @xmath79 caii system ( @xmath85kms@xmath16 ) , while the mgii bal trough begins at @xmath80 , the redshift of the other caii system ( @xmath86kms@xmath16 ) .
note that sdss 0300 + 0048 is a binary quasar with sdss j025959.69 + 004813.5 , a non - bal quasar located 195 away at @xmath87 ( @xmath88kms@xmath16 ) .
the smooth troughs in sdss 1154 + 0300 ( figure [ f_1154 ] ) make its redshift difficult to pin down .
we adopt @xmath89 from various emission features .
the absorption probably begins at this redshift ( e.g. , in aliii@xmath621854,1862 ) , but does not reach its full depth until @xmath90 .
rest wavelengths at the latter redshift are plotted along the top axis of figure [ f_1154 ] .
the more rapid onset of absorption in aliii compared to mgii suggests that the bal region may cover the aliii broad line region but not the mgii broad line region .
sdss 0819 + 4209 ( figure [ f_abruptbals]a ) was discovered in a search for @xmath91 quasars among @xmath12-dropouts in sdss images ( fan et al .
the sharp drop due to mgii absorption mimicks the onset of the ly@xmath22 forest at high redshift .
the spectrum was obtained with esi ( epps & miller 1998 ) at keck ii on ut 19 mar 2001 .
we adopt @xmath92 from unresolved mgi absorption , accompanied by broader mgii absorption , located 2180@xmath3780kms@xmath16 longward of the onset of the absorption troughs at @xmath93 .
sdss 1730 + 5850 ( figure [ f_abruptbals]b ) has @xmath94 and shows no flux , within the errors , below alii@xmath551670 .
it may have very extensive hei@xmath553188 absorption starting just shortward of @xmath95 , the onset redshift of the mgii and mgi bal troughs , though this needs confirmation given the strong telluric absorption at those observed wavelengths .
this object was discovered using the double imaging spectrograph ( dis ) at the apo 3.5 m on ut 27 may 2000 during exploratory spectroscopy of sdss objects with odd colors .
spectra were later obtained twice by the sdss , on ut 23 aug 2000 and 18 apr 2001 .
no significant variability was detected between those observations ( not surprising given the low snr of the spectra ) .
the coadded sdss spectrum is a factor of @xmath02.25 fainter than the discovery spectrum .
this discrepancy , while large , could be due to the uncertainties in the sdss fluxing and in comparing slit and fiber spectra .
the scaled sdss spectrum shows good agreement with the discovery spectrum at @xmath96 , but the ` peaks ' at 1900 , 2100 and 2500 are stronger by a factor of 1.42 .
that is , the absorption at those wavelengths appears weaker than in the discovery spectrum .
nonetheless , the uncertainties are so large that we do not feel this is a firm detection of variability , though it does suggest that careful monitoring of this object might be worthwhile . in figure [ f_abruptbals ]
we have therefore summed the scaled discovery spectrum and the coadded , smoothed sdss spectrum to achieve the best snr and wavelength coverage .
sdss 0437@xmath180045 was selected from early sdss images by xf and mas as having unusual colors ( fan et al .
optical spectra were obtained at the apo 3.5 m using dis on ut 22 mar 1999 , 29 dec 1999 , and 04 jan 2000 , at keck ii using the low resolution imaging spectrograph ( lris ; oke et al .
1995 ) on ut 15 oct 1999 , and at the canada - france - hawaii telescope ( cfht ) using the subarcsecond imaging spectrograph ( sis ; crampton et al .
1992 ) on ut 27 jan 2001 .
standard iraf reduction procedures were used to obtain flux - calibrated spectra . near infrared spectra covering 1.45 - 2.09 @xmath97 m at resolution 0.0025 @xmath97 m and 1.0 - 1.32 @xmath97 m at resolution 0.0050
@xmath97 m were obtained using the cooled grating spectrometer ( cgs4 ; mountain et al .
1990 ) at the united kingdom infrared telescope ( ukirt ) on ut 27 and 30 sep 1999 respectively .
the longer wavelength spectrum was obtained under non - photometric conditions .
the data were reduced using the starlink figaro package .
near - ir ( nir ) photometry obtained by hwr at calar alto in nov .
1999 yields @xmath98 , @xmath99 and @xmath100 , making @xmath101 somewhat bluer than average for a quasar .
the spectra at 1.1 - 1.312 and 1.5 - 1.75 were normalized to the fluxes expected from the @xmath26 and @xmath102 photometry , respectively .
a further correction of 1.33 was applied to the longer - wavelength spectrum to make the spectral slope continuous between @xmath26 and @xmath102 .
the true slope is probably bluer than shown , since the more trustworthy photometry indicates the object is bluer than the the long - wavelength spectrum is .
figure [ f_weird2 ] shows the combined optical ( keck ) and nir ( ukirt ) spectrum of sdss 0437@xmath180045 , which finally enabled us to identify the object after being stumped by it for some time .
it is a strongly absorbed quasar with troughs that reach peak depth at @xmath103 .
there is evidence for aliii absorption at @xmath104 . since the lack of abrupt long - wavelength edges to the absorption troughs in
the optical suggests that the systemic @xmath75 is higher than the peak absorption @xmath75 , we adopt @xmath104 as our systemic redshift . in figure[f_weird2 ] we plot emission lines at this redshift and absorption lines at the peak absorption redshift .
there seems to be considerable absorption from neutral gas , namely several fei lines , mgi@xmath552852 , and mgi@xmath552026 , the latter possibly blended with znii .
there is no evidence for caii absorption , but we plot its wavelengths for reference . as for emission ,
civ appears absent because it is completely eaten away by feii multiplets uv44,45,46 .
note that the lines we identified as [ oiii ] at @xmath105 in ( 2001 ) are at the wavelengths of night sky lines , and are almost certainly not real .
this object may have h@xmath1 absorption nearly 10@xmath106kms@xmath16 wide , with rest - frame ew@xmath0100 .
h@xmath1 absorption in agn has previously been seen only in ngc 4151 ( anderson & kraft 1969 ; sergeev et al .
1999 ) , and there with @xmath781000kms@xmath16 width and @xmath783 rest - frame ew . however , the snr is not high enough to be sure this dip in the continuum is real .
even if so , it may just be a gap between h@xmath1 and feiiopt37,38 emission ( e.g. , figure 4 of 1994 ; see also the spectrum of fbqs 1408 + 3054 in [ gradpc ] ) .
the expected wavelengths of emission from this and three other strong feii emission multiplets are plotted in figure[f_weird2 ] .
the features @xmath075 to the red of each of them argue that sdss 0437@xmath180045 has strong feii emission at a slightly higher systemic redshift than we have assumed , and not an h@xmath1 bal trough .
a better spectrum is needed .
the absorption shortward of 2500 in this object has varied with an unusually high amplitude and rate of change for bal quasars .
figure[f_w2time]a shows the optical spectrum of sdss 0437@xmath180045 at four different epochs .
the spectra have been normalized at 7000@xmath187590 , where the snr is highest , to match the cfht spectrum . for reference ,
figure [ f_w2time]b compares the cfht and keck spectra without normalization , illustrating the typical variation in absolute flux among our spectra . our normalization accounts for the uncertainty in the absolute flux calibration of these narrow slit observations .
the relative wavelength - dependent flux calibrations are trustworthy ; for example , the normalized keck and second - epoch apo spectra , taken within 20 rest - frame days of each other , agree completely within the errors across the wavelength range 4500@xmath189000 .
this also reassures us that the variability is not due to fluctuations in system sensitivities or flatfields . to calculate the variation of the absorption strength in sdss 0437@xmath180045
, we assume the true continuum is flat in @xmath49 at @xmath107 ergs cm@xmath29s@xmath16@xmath16 .
then , relative to the 7000@xmath187590 region , in the 90 rest - frame days between the keck and cfht spectra epochs the absorption weakened by 5@xmath371% at 5200@xmath185600 ( high - velocity civ ) and strengthened by 8@xmath371% at 5900@xmath187000 ( where narrow troughs of feii , alii and aliii are visible ) and [email protected]% at 7590@xmath188150 ( mgi and feii , plus znii , crii and feiii ? ) .
this last spectral region contains the strong telluric o@xmath108 absorption band at 7590@xmath187700 , but the increase in absorption is larger than the telluric correction applied and the region of apparently increased absorption is wider than the absorption band .
this increase in the relative absorption at 7590@xmath188150 is somewhat less than the 15@xmath373% increase in this region in the 54 rest - frame days between the discovery and keck spectra ( figure[f_w2time]a ) .
the variable absorption in sdss 0437@xmath180045 is discussed further in [ var ] .
sdss has discovered a number of heavily reddened bal quasars . after a short discussion of how we determine the reddening in these objects , we present two reddened mini - bals with strong feii emission , then several extremely reddened objects with no strong emission , and finally a bright reddened felobal .
the implications of all these objects are discussed in [ impred ] .
since the ` typical ' quasar has a very blue spectrum , reddened quasars are easy to identify .
determining the amount of reddening is more difficult .
we assume all reddening occurs at the quasar redshift with a small magellanic cloud ( smc ) extinction curve ( prevot et al .
1984 ) and @xmath109 ( see below ) .
we deredden the spectrum until the continuum slope matches that of the composite sdss quasar of ( 2001 ) .
the value of the color excess @xmath2 needed to achieve this match is our estimated reddening .
the uncertainties on @xmath2 for each object denote the range for which an acceptable match can be found .
we use the smc extinction curve because the 2200 bump present in the lmc and milky way extinction curves has never been detected from dust around quasars ( e.g. , pitman , clayton , & gordon 2000 ) , although it has been detected from dust in intervening mgii systems ( malhotra 1997 ; cohen et al . 1999 ) .
the smc curve was also used by & foltz ( 1992 ) and ( 2001b ) , both of whom found @xmath2=0.1 for a ` typical ' lobal spectrum .
the other commonly used extinction curve is the calzetti formula ( calzetti , kinney , & storchi - bergmann 1994 ) .
this formula was derived for active star formation regions and empirically incorporates the ` selective attenuation ' effects of dust , including extinction , scattering , and geometrical dust distribution effects .
the calzetti extinction curve does not have a 2200 bump , but like the lmc and mw curves it is much greyer ( less steep ) than the smc extinction curve .
our use of the smc curve instead of the calzetti curve is conservative in the sense that it requires a lower @xmath2 ( and thus lower extinction ) for a given observed ultraviolet slope .
two more caveats to our dereddening procedure are needed .
first , we assume a single value of @xmath2 over all sightlines to the emission regions . a range of @xmath2 ( e.g. , hines & wills 1995 ) or a wavelength - dependent contribution from scattered light ( brotherton et al .
2001a ) will complicate the interpretation of our derived single @xmath2 .
however , sometimes we can tell when these effects are important ( e.g. , sdss 0342 + 0045 ; see below ) .
second , if an accretion disk is producing the observed emission and the amount of extinction is correlated with the inclination of the disk , then dereddening to match the sdss composite will introduce a systematic error in the derived @xmath2 since the intrinsic continuum of the disk is likely to be a function of viewing angle ( e.g. , hubeny et al .
2000 ) .
sdss 1453@xmath1100029 was identified by xf and mas as having unusual colors in early sdss imaging ( fan et al .
a discovery spectrum was obtained at the apo 3.5 m , and a 1200s followup spectrum at keck ii on ut 05 apr 2000 ( figure [ f_weird1]a ) .
the object was observed numerous times as a quasar candidate in the sdss spectroscopic survey .
we adopt the redshift @xmath111 determined by the pipeline from one of these observations ( the other yielded no redshift ) , since it agrees well with the keck spectrum .
the redshift comes from narrow mgii emission and weak ciii ] emission .
the object is similar to q 2359@xmath181241 ( brotherton et al .
2001a ) ; both are reddened quasars with narrow mgii emission and absorption atop a complex broader structure .
both objects also show hei@xmath553188 and hei@xmath553889 in absorption ( arav et al .
sdss 1453 + 0029 has bi=0 , but ai(mgii)=253@xmath18477kms@xmath16 ( the lower value is from the keck spectrum and the higher value from the coadded sdss spectrum ) , with @xmath112kms@xmath16 . a similar object is sdss 0127 + 0114 , which has aliii , mgii , hei@xmath553188 and hei@xmath553889 absorption and narrow ciii ] , [ oii ] and [ neiii ] emission , as well as broader mgii emission , at @xmath113 ( figure [ f_weird1]b ) .
sdss 0127 + 0114 has zero balnicity , but the presence of hei argues that the absorption is a mini - bal trough and not simply an associated narrow absorption line system .
we estimate smc reddenings of @xmath2=@xmath114 for sdss 1453 + 0029 and @xmath2=@xmath115 for sdss 0127 + 0114 .
these dereddened spectra are plotted atop the normalized composite sdss spectrum in figure [ f_w1dered ] , along with the observed spectrum of the extreme feii - emitting quasar sdss [email protected] ( schneider et al .
sdss 0127 + 0114 is clearly a normal - to - strong feii emitter , while sdss 1453 + 0029 is intrinsically an extreme feii - emitting quasar similar to sdss [email protected] or q 2226@xmath183905 ( graham et al .
what looks like a detached mgii bal trough at 2630 - 2700 in the observed spectrum of sdss 1453 + 0029 is actually the gap between mgii+feiiuv62,63 emission and feiiuv1,64 emission shortward of @xmath02630 .
similarly , gaps between mgii and feiiuv78 emission and feiiuv78 and feiiopt6,7 produce apparent troughs at 2900 and 3000 - 3150 , respectively .
feii emission this strong remains a challenge for quasar models ( e.g. , sigut & pradhan 1998 ) .
sdss has discovered several extremely reddened bal quasars whose emission lines in the observed optical are very weak or absent , somewhat reminiscent of hawaii 167 ( cowie et al .
1994 ) . for these objects
we assume that the onset redshifts of the absorption troughs are the systemic redshifts ; where present , the weak ciii ] lines are consistent with these redshifts .
we discuss the observed spectral slopes of these objects in terms of @xmath116 , where @xmath117 . for reference ,
the sdss quasar composite has @xmath118 , and ( 2002 ) adopted @xmath119 as their working definition of a red quasar .
[ [ sdss0947 ] ] sdss 0947 + 6205 + + + + + + + + + + + + + + sdss 0947 + 6205 ( figure [ f_redbals12]a ) is at @xmath120 from feii and mgii absorption .
the mgii absorption trough spans some 3700@xmath37200kms@xmath16 , but the bi=0 since civ region is too noisy to measure and the mgii absorption begins at the systemic redshift .
the top axis of figure [ f_redbals12]a gives the rest wavelength at the peak absorption redshift of @xmath121 ( detachment @xmath122kms@xmath16 ) .
alii , aliii , and civ absorption are also present , detached by 1500@xmath3770kms@xmath16 and with a narrower velocity width of 3700@xmath37200kms@xmath16 .
we measure a spectral slope of @xmath123 , and dereddening by @[email protected] brings the spectrum into fair agreement with the sdss composite . however , there is some suggestion that shortward of aliii the object s spectral slope may steepen and the required reddening may increase , or that the reddening curve is different from that assumed , as we discuss below .
[ [ sdss1324 ] ] sdss
1324@xmath180217 + + + + + + + + + + + + + + + + + + + + + + + + + + + + this object ( figure [ f_redbals12]b ) is at @xmath124 from broad mgii , feii , and alii absorption ; it may also have alii and civ absorption .
there is also broad aliii , mgii , and probably feiiiuv34,48 and weak feii absorption at @xmath125 , an outflow velocity of 13230@xmath37130kms@xmath16 ; nonetheless , the object has bi=0 .
we measure a spectral slope of @xmath126 , but the slope appears to steepen shortward of aliii . dereddening the spectrum to match the sdss composite confirms this steepening , which could be intrinsic or due to an extinction curve different from that of the smc .
we discuss this in detail in [ impred ] , but here we simply quote the values of @xmath2 needed to bring the object spectrum into agreement with the composite at different wavelengths , using the smc extinction curve .
a reddening of @[email protected] is required at 2000@xmath182800 , but @[email protected] is required at 1500@xmath182000 .
[ [ sdss1456 ] ] sdss 1456 + 0114 + +
+ + + + + + + + + + + + for this extremely reddened first - detected bal quasar ( figure [ f_redbals34]a ) we adopt @xmath127 .
it shows broad ciii ] emission , weak narrow ly@xmath22 emission , and broad civ , aliii and feii absorption along with narrow sii , znii , crii , fei and si absorption at the deepest part of the trough , @xmath128 ( the top axis of figure [ f_redbals34]a gives rest wavelengths at this redshift ) .
we measure a spectral slope of @xmath129 , and there is no evidence for any deviation from that slope
. dereddening by @[email protected] is required to match the sdss composite .
[ [ sdss0834 ] ] sdss 0834 + 5112 + +
+ + + + + + + + + + + + this object ( figure [ f_redbals34]b ) is a first source at @xmath130 from mgii , feii , aliii , and weak civ absorption .
the object has two detached absorption troughs : one at @xmath131 in broad civ and mgii , and one at @xmath132 in narrow mgii , feii , aliii , civ and siiv .
this latter system could be intervening ; the snr is not high enough to tell if its civ absorption is broad .
this object can not be adequately fit by a single power law .
the spectrum steepens from @xmath126 at 2450@xmath182725 to @xmath133 at 2000@xmath182450 and again to @xmath134 at 1550@xmath182000 . because of this steepening of the continuum , dereddening by @[email protected] is required to match the sdss composite at 2000@xmath182800 , but @[email protected] is required at 1500@xmath182000 ( figure[comp0834p5112sm7_smc0.3 ] ) .
[ [ sdss0342 ] ] sdss 0342 + 0045 + + + + + + + + + + + + + + we adopt @xmath135 for this felobal ( figure [ f_0342p0045 ] ) from numerous absorption lines .
the spectrum shown was obtained at keck ii using lris on ut 15 oct 1999 .
there is weak civ absorption and maybe weak ly@xmath22+nv emission , but the spectrum is dominated by strong feii absorption ( multiplets uv1,2,3 at least ) .
it is obvious from comparison to figures [ f_redbals12 ] and [ f_redbals34 ] that this is a somewhat different sort of extremely reddened bal quasar . instead of the spectrum steepening at shorter wavelengths
, it flattens out .
this is what is expected for a range of reddenings along the sightlines to the emission regions : the heavily reddened sightlines contribute only at longer wavelengths , while less reddened sightlines dominate at shorter wavelengths .
we find that dereddening by @[email protected] is required to match the sdss composite between 2050@xmath182550 , while only @[email protected] is required between 1250@xmath181850 .
the evidence in several objects in the previous section for a reddening curve steeper than that of the smc is confirmed in dramatic fashion by this object .
sdss 0318@xmath180600 ( figure [ f_0318]a ) has @xmath136 from cross - correlation with the sdss composite quasar of ( 2001 ) .
it has flux down to at least 1300 rest frame and shows absorption from a large number of transitions , few of which appear to reach zero flux .
it is similar to fbqs 1044 + 3656 ( de kool et al .
2001 ) , except that mgi@xmath552852 absorption is only tentatively detected , along with hei@xmath552945 .
the long wavelength edges of the bal troughs are at @xmath137 ( a detachment velocity of 2650@xmath37160kms@xmath16 ) , with the strongest absorption at @xmath138 .
the civ trough extends to an outflow velocity only 3000kms@xmath16 higher than the mgii trough .
the absorption just longward of feii@xmath552600 from the highest - redshift system ( @xmath139 ) shows that excited - state feii absorption is present .
multiplets uv1 , uv2 , and uv3 can be firmly identified , but more may be present .
the absorption just longward of alii , at @xmath05000@xmath185100 observed , is probably due to a combination of niii and feiiuv38 .
dereddening the spectrum by @[email protected] using the smc extinction curve brings its slope into agreement with that of the sdss composite quasar at 2000 - 3000 rest frame , but @[email protected] is required to bring the slope at 1250 - 2000 into agreement with the composite ( figure [ f_0318]b ) . note that sdss 0318@xmath180600 is detected by 2mass , with @xmath140 , @xmath141 and @xmath142 .
its near - ir colors are _ bluer _ than those of most quasars , indicating that the above @xmath2 may be somewhat underestimated . the implications of our observations of sdss 0318@xmath180600 and all the other objects presented earlier in this section are discussed in [ impred ] .
we end this section on heavily reddened bal quasars with a discussion of an object possibly similar to sdss 0318@xmath180600 . in sdss 0338 + 0056 at @xmath143 ,
the spectrum declines precipitously shortward of ciii ] ( figure[f_0338 ] ) .
this decline is confirmed by the object s @xmath144 color . the spectrum between ciii ] and
mgii seems to be a blend of feii emission and mgii and feii absorption , but the object is not yet definitively understood .
we outline two possible interpretations here .
the object may be a strong feii - emitting quasar which is reddened with an extinction curve steeper than that of the smc . in this interpretation ,
the approximate continuum is given by the dashed line in figure[f_0338 ] and the apparent bal troughs are in fact just gaps between strong feii emission multiplets .
the smoothness of the spectrum from 1750@xmath181900 is hard to explain in this model , since there are feii multiplets in that wavelength range .
however , the object does resemble sdss 0318@xmath180600 ( [ sdss0318 ] ) without the bal troughs , so we consider this interpretation is plausible . alternatively , the object may be a reddened bal quasar with strong aliii+feiiiuv34 absorption . in this interpretation ,
the continuum shortward of mgii is given by the dot - dashed line in figure[f_0338 ] .
there is a good match between the putative troughs of feiiiuv34 , feiiuv1 , feiiuv3 and mgii .
also , feiiiuv34 absorption would be stronger than feiiiuv48 absorption , which is normal
. however , alii absorption is weak at best and the start of a civ trough is not clearly detected .
if better snr in the blue confirms the lack of a civ bal , then this bal quasar hypothesis would be ruled out .
even if the bal hypothesis is correct , the apparently wider trough in aliii than in mgii would require explanation .
it could be due to unusually strong niii absorption at the high - velocity end of the apparent aliii trough , high - velocity mgii absorption masked by feii emission , or an overestimated continuum level at @xmath191750 due to strong heii+oiii emission . in the latter case
, the bal quasar model continuum should match the reddened strong - feii - emitter model continuum at @xmath191750 .
the resulting very steep drop in the continuum at @xmath191900 would require either an extinction curve steeper than that of the smc or a red continuum with the same origin as the red continua of the mystery objects of [ gradual ] , whatever that origin is .
sdss 2215@xmath180045 ( figure [ f_2215 ] ) is a reddened lobal with detached mgii@xmath622796,2803 , aliii@xmath621854,1862 , alii@xmath551670 , and feiii absorption .
our redshift @xmath145 is set by narrow associated mgii absorption seen atop a weak broad mgii line .
sdss 2215@xmath180045 is detected by 2mass , with @xmath146 , @xmath147 and @xmath148 .
its near - ir colors are bluer than most quasars .
this is consistent with the spectrum longward of 3000 , which is as blue as the sdss composite .
the spectrum shortward of 3000 , however , appears to be reddened by @xmath2@xmath149 compared to the sdss composite .
this will be a lower limit to the reddening if sdss 2215@xmath180045 is indeed bluer than the average quasar . by comparison to sdss 1723 + 5553 ( figure [ f_scalloped3 ] , bottom left corner ) , we initially identified the strong trough at @xmath150@xmath04900 as crii@xmath55@xmath55@xmath552056,2062,2066 .
however , the implied relative abundance of cr is implausible , and the expected corresponding znii is missing .
this absorption is in fact due to feiiiuv48 @xmath55@[email protected],2068.90,2079.65 ( ep=5.08ev ) .
there is also feiiiuv34 @xmath55@[email protected],1914.06,1926.30 ( ep=3.73ev ) absorption longward of aliii , at @xmath150@xmath04500 .
however , feii absorption is weak at best .
no feii absorption troughs are detectable to a limit of 10% of the strength of the feiii troughs .
in addition , mgii absorption is probably much weaker than it appears .
there are gaps between feii emission complexes at @xmath02670 ( rest - frame ) , the same wavelength as the detached mgii absorption , and at @xmath03100 ( e.g. , figure [ f_w1dered ] ) .
given the strength of the feii emission around the 3100 gap , most of what appears to be detached mgii absorption in sdss 2215@xmath180045 could be due to the 2670 gap .
however , the presence of other absorption troughs at this redshift argues that some mgii absorption is present .
both feii emission modeling and spectropolarimetry would be useful in untangling this object s absorption troughs from its feii emission . in figure[f_2215vplots5 ]
we plot the normalized spectrum of sdss 2215@xmath180045 around the troughs of mgii , feiiiuv48 , feiiiuv34 , aliii , and alii .
all troughs were normalized using the continuum fit shown as the dot - dashed line in figure [ f_2215 ] , though as discussed above , the mgii trough is confused with complex feii emission .
the strength of the mgii trough could well be less than shown , and its velocity structure is also untrustworthy .
the troughs are plotted in terms of blueshifted velocity from systemic ; for multiplets , the velocity is for the longest wavelength line .
the dot - dashed vertical lines show the wavelengths of every line of each multiplet at 14300kms@xmath16 , the central velocity of the alii absorption . within the uncertainties ,
the troughs appear to have the same velocities for their peak absorption , and all troughs except for alii are consistent with having the same starting and ending velocities ( 6000 and 18000kms@xmath16 , respectively ) .
the absorption troughs are thus unusual for a lobal in that they are strongest near the high - velocity end rather than near the low - velocity end . at the velocity of peak absorption the normalized depths of the feiiiuv48 and aliii lines
are the same , within the uncertainties .
this suggests that they share the same partial covering factor .
the alii and feiiiuv34 troughs are not as deep ; this is probably because they are not saturated , rather than the depth being a reflection of covering fraction , since feiiiuv34 and feiiiuv48 absorption should have the same covering factor .
finally , the relative strengths of uv48 and uv34 absorption are reversed in this object compared to other bal quasars with feiii ( e.g. , sdss 1723 + 5553 , [ hiscal ] ) .
we discuss the implications of this in [ impfe3 ] .
we have found several other candidate strong feiii absorbing bal quasars besides sdss 2215@xmath180045 , though none with feiii absorption as strong relative to mgii as in that object .
the criteria we use for selecting such candidates is the presence of a trough at @xmath02070 which is stronger than feiiuv1,2,3 troughs near 2400 and 2600 .
when such feii troughs are present with strength comparable to or greater than the 2070 trough , crii is a more likely identification for the latter than feiii .
we note in passing that pc 0227 + 0057 ( schneider , schmidt , & gunn 1999 ) has been identified as a similar object at @xmath151 .
it has weak feii absorption , but its 2070 trough probably includes a contribution from feiii .
the trough is too strong relative to the feii troughs to be due solely to crii .
[ [ sdss1214 ] ] sdss 1214@xmath180001 + + + + + + + + + + + + + + + + + + + + + + + + + + + + sdss 1214@xmath180001 ( figure[f_sdss1214 ] ) is at @xmath152 from associated mgii absorption and [ oii ] emission ( the latter is real , despite being near a night sky line ) .
the peak of the broad mgii emission is shortward of this redshift .
there is a detached mgii trough , but no matching feii troughs .
thus the 2070 trough is likely to be feiiiuv48 .
this claim is bolstered by the apparent start of a feiiiuv34 trough at the blue edge of the spectrum , at a redshift which matches the start of both the feiiiuv48 and mgii troughs .
higher snr data extended further to the blue are needed to study this object further .
[ [ sdss0149 ] ] sdss 0149@xmath180114 + + + + + + + + + + + + + + + + + + + + + + + + + + + + sdss 0149@xmath180114 at @xmath153 ( figure [ crii]a ) has low snr but does appear to have a narrow feiiiuv48 trough as well as a broader mgii trough and weak feii uv1,2,3 absorption .
higher snr data are needed to confirm the strong feiiiuv48 line and to measure the strength of feiiiuv34 .
[ [ sdss0810 ] ] sdss 0810 + 4806 + + + + + + + + + + + + + + sdss 0810 + 4806 at @xmath154 ( figure [ crii]b ) may have weak feii uv1,2,3 absorption , but the 2070 trough is much stronger than any such absorption , meaning that it is almost certainly feiii instead of crii .
however , higher snr data are needed to say anything beyond that .
the sdss has discovered two objects to date with strange but very similar continuum shapes that include dropoffs shortward of mgii .
both are first sources with fluxes of a few mjy .
sdss 0105@xmath180033 has [ oii]@xmath623727,3729 emission at @xmath155 ( figure [ gradualbals]a ) .
no other narrow emission lines are present between 1800aa and 3900 ( rest - frame ) , but there may be broad emission features near 4000 and 4200 , possibly due to feii .
there is also unresolved associated mgii absorption at @xmath156 .
given that the [ oii ] redshift agrees with this within its errors , we adopt the more accurate mgii redshift as systemic .
sdss 0105@xmath180033 is detected by 2mass , with @xmath157 , @xmath158 and @xmath159 .
its near - ir colors are typical for quasars .
sdss 2204 + 0031 does not show any narrow emission , but it has associated mgii absorption at @xmath160 ( figure [ gradualbals]b ) .
however , comparison of its spectrum with that of sdss 0105@xmath180033 shows that the various continuum features do not match up if this mgii redshift is assumed to be the systemic redshift .
cross - correlation of the two spectra yields a redshift @xmath161 for sdss 2204 + 0031 .
the snr of the cross - correlation peak is @xmath162 , which is low .
however , given the similarity of the continuum features in the two objects , we adopt the cross - correlation redshift for sdss 2204 + 0031 .
this places the mgii absorption 2260@xmath37120kms@xmath16 shortward of systemic .
longward of about 3200 rest frame , the continuum in both objects is blue and mostly featureless . as seen in figure [ gradualbals ] , moving shortward in the rest frame , both observed ( unreddened ) spectra show a peak near 3200 , a local mininum near 3000 , a relatively flat ( in @xmath163 ) region with associated mgii absorption , a slight drop near 2600 - 2625 , another drop near 2500 ( stronger in sdss 0105@xmath180033 ) , and finally a continuum which decreases slowly to shorter wavelengths down to at least 1750 .
the dropoffs shortward of the mgii absorption appear too steep to be due to reddening and are not due to obvious bal troughs . the only objects with similar spectra that we are aware of are fbqs 1503 + 2330 ( @xmath164 ) and fbqs 1055 + 3124 ( @xmath165 ) both of which have narrow mgii absorption and strong broad feii and balmer line emission at the same redshift ( white et al .
fbqs 1503 + 2330 has no obvious mgii emission , while fbqs 1055 + 3124 has a spectrum near mgii closely resembling that of sdss 0105@xmath180033 . just like our two sdss objects
, these objects have dropoffs shortward of mgii which are too abrupt to be caused by reddening and are not obviously due to bal troughs .
since these features are at the blue ends of the spectra , however , they could be due to detached mgii bal troughs or to flux calibration errors ( cf . the difference between the spectra of fbqs 1044 + 3656 in 2000 and 2001 ) .
we consider possible explanations for the spectra of these objects in [ gradexp ] .
having presented five categories of unusual bal quasars discovered to date in the sdss , we now discuss their implications for models of bal outflows and quasars in general .
we do so in a somewhat unorthodox reverse order so that our lengthy discussion of quasars with longward - of - systemic absorption comes last .
sdss 0105@xmath180033 and sdss 2204 + 0031 ( [ gradual ] ) are not simply reddened normal quasars , since they lack obvious emission lines .
however , reddened bal quasars can have weak emission lines ( [ extred ] ) , and these high - redshift radio sources must be agn since they are more luminous than any galaxy .
since reddening seems to be present , we attempt to account for it and to see if reddening of unusual agn can explain the spectra .
to estimate the reddening in these objects , we followed the procedure outlined in [ getred ] , except that we interpolated over all strong emission lines in the composite sdss quasar before comparing to it . even before dereddening ,
both objects are roughly as blue as the composite when all three spectra are measured at rest wavelengths 3000@xmath183900 .
therefore , if reddening is present , these objects must be intrinsically bluer than the average quasar .
values , quasars with bluer intrinsic uv - optical colors and lower @xmath2 will be overrepresented at brighter fluxes . ] by considering the entire 1800 - 3900 region , we estimate minimum reddenings of @xmath2@xmath166 for sdss 0105@xmath180033 and @xmath2@xmath167 for sdss 2204 + 0031 using the smc extinction curve at the quasar redshift .
the dereddened spectra are compared to the composite spectrum in figure [ gradered ] .
some quasars have very weak mgii emission but normal feii emission , and some have very strong feii emission ( e.g. , [ weird1 ] ) .
are the strange continua of sdss 0105@xmath180033 and sdss 2204 + 0031 consistent with reddened versions of such quasars ?
some feii emission may be present in these two objects the peak near 3200 seen in many quasar spectra is usually ascribed to feiiopt6,7 and the local minimum near 3000 to a gap between feii multiplets uv78 and opt6,7+[oiii ] ( vanden berk et al .
2001 ) but their spectra shortward of mgii do not show emission at the wavelengths of feii multiplets seen in normal and strong feii emitters ( compare figure [ gradered ] to figure [ f_weird1]b ) .
thus , while reddening may be present in these objects , it can not fully explain their unusual spectra .
the two fbqs objects similar to these two sdss objects ( [ gradual ] ) have strong feii emission .
since strong feii emission and the presence of bal troughs are correlated ( boroson & meyers 1992 ) , this suggests that all four objects are indeed bal quasars .
if so , shorter wavelength spectra of the two fbqs objects should reveal detached but otherwise normal bal troughs . for the two sdss objects a more unusual type of bal trough is required , as follows , but even that does not provide a very satisfactory explanation .
we can achieve barely plausible fits for the two sdss objects as bal quasars with mgii , feii@xmath552750 , and feii@xmath552600 troughs @xmath4520,000kms@xmath16 wide and detached by @xmath012300kms@xmath16 . however , feii@xmath552400 seems weaker than expected .
moreover , most bal quasars have troughs which are saturated in most transitions even if there is only partially covering of the continuum source ( arav et al . 2001b ) .
thus the absorption does not appear twice as strong when two troughs overlap ( i.e. , when two species have different outflow velocities such that they produce absorption at the same observed wavelength ) .
in contrast , explaining these objects as bal quasars requires absorption that increases in strength when troughs overlap .
unsaturated absorption would do this , but the putative feii@xmath552600 absorption here is much too strong relative to mgii for the troughs to be unsaturated . the only other way to produce absorption which increases in strength when troughs overlap is to have partial covering of _ different _ regions of the continuum source as a function of velocity .
partial covering of the _ same _ region of the source with velocity is what is seen or assumed in most bal quasars ( e.g. , 3.3 of de kool et al .
the only object known to definitely exhibit spatially distinct velocity - dependent partial covering is fbqs 1408 + 3054 ( figure [ f_1408 ] ; spectrum from 2000 ) .
the feiiuv1 trough in this object has smaller partial covering than the mgii trough .
however , where the high velocity end of the feiiuv1 trough overlaps with the low velocity end of the feiiuv2,3 trough , at 4300 observed , the absorption increases in depth .
( contrast this with the top panel of figure[f_scalloped3 ] , where all feii troughs have the same apparent partial covering even when they overlap . ) different partial covering of the same source region can not explain this , since feii uv1 , uv2 , and uv3 absorption arise from the same term of feii .
nor can the gas be optically thin since feii absorption would not be nearly as strong relative to mgii in that case .
the feii gas must partially cover different regions of the continuum source as a function of velocity .
spatially distinct velocity - dependent partial covering may not be all that rare , just difficult to recognize .
unless the spatial regions covered change rapidly with velocity , such partial covering would be obvious only when two troughs overlap at substantially different velocities .
troughs that wide are uncommon and not very well studied .
nonetheless , given that even with this type of partial covering the fits for these objects as bal quasars are barely plausible , we consider other possible explanations for these objects .
the region from 25003200 in these objects could be a region of continuum emission plus extremely broad ` double - peaked ' or ` double - shouldered ' mgii emission produced by an accretion disk ( e.g. , arp102b , cf .
figure 1 of halpern et al . 1996 ) .
if so , the full width at zero intensity ( fwzi ) in both objects would be about 40,000kms@xmath16 , twice the fwzi of the double - shouldered lines in arp 102b .
( 1996 ) model arp 102b as an accretion disk inclined by 32 to the line of sight .
the same disk viewed edge - on would have a fwzi 1.9 times larger , so a fwzi of 40,000kms@xmath16 is possible . the observed
reddening is probably consistent with an edge - on line of sight .
however , as seen in figure [ gradered ] , to explain the dereddened continuum shape shortward of mgii , the double - shouldered emission line hypothesis may require shallow , smooth absorption troughs from @xmath02050 - 2500 rest frame in sdss 0105@xmath180033 and @xmath01900 - 2500 rest frame in sdss 2204 + 0031 .
there are feii transitions in this wavelength range , but they should be accompanied by mgii absorption .
we therefore must identify the start of the trough at 2500 with mgii absorption detached by 35,000kms@xmath16 .
the absorption must also reach an outflow velocity of @xmath4653,000kms@xmath16 so that the mgii and feiiuv1,2,3 troughs all overlap to produce a single smooth trough .
both the detachment and peak outflow velocities are very large for low - ionization bal troughs , but not unprecedented ( [ abrupt ] ) .
alternatively , the difference between the dereddened continuum shapes and the composite may be due to the absence of feii emission in these two objects .
similarly , adding a small 2200 bump to the smc extinction curve might eliminate the need for these broad , shallow absorption troughs , and the drop near 2500 could be explained as a moderately broad detached mgii trough ( cf .
sdss 2215@xmath180045 ; [ sdss2215 ] ) with little or no accompanying feii absorption .
our two sdss objects are first sources , but radio - quiet .
thus the peaks in the spectra near mgii in these objects can not be due to the turnover of a red synchrotron component ( e.g. , francis , whiting , & webster 2000 ) .
these objects are not supernovae , as supernovae have redder colors at these rest wavelengths .
they are not galaxies at the redshifts of the mgii absorption which are gravitationally lensing higher - redshift quasars , since there are no signs of emission lines or ly@xmath22 forest absorption from such quasars .
neither object shows any sign in its images or spectrum of being a superimposition of two quasars or a quasar and a star or galaxy .
one very speculative possible explanation is emission from an accretion disk with a gap in the regions hot enough to produce significant emission at @xmath168 .
such a gap is unlikely to be long - lived , so future observations of these objects could test this hypothesis .
one last possibility is a reddened version of pg 1407 + 265 ( mcdowell et al .
1995 ) , which has very weak emission lines shortward of mgii , and emission lines increasingly blueshifted with ionization potential , up to a maximum of 10,000kms@xmath16 in civ relative to h@xmath22 . reddened quasars with weak mgii and weak , blueshifted ciii
] might explain these objects , though their detailed continuum structure would still be puzzling . none of the three viable explanations for these objects bal quasars with spatially distinct partial covering , double - shouldered emission line objects , or reddened pg 1407 + 265 analogs are particularly satisfactory .
the spatially distinct partial covering hypothesis predicts these objects should have other properties similar to bal quasars : e.g. , weak x - ray emission , strong feii and weak [ oiii ] emission , and high polarization , especially in the bal troughs .
the double - shouldered hypothesis can be tested by studying the line profiles of h@xmath1 or h@xmath22 in the near - ir or of civ in the atmospheric uv ; our spectra are unfortunately too noisy to determine if there is double - shouldered h@xmath68 emission in either object .
the reddened pg 1407 + 265 hypothesis can be tested with high - snr spectra of ciii ] and civ , as well as h@xmath1 and h@xmath22 .
finally , note that it is possible that both the [ oii ] emission and mgii absorption in sdss 0105@xmath180033 are from an intervening galaxy .
if this is the case then the redshifts of _ both _ objects are unknown , since our redshift for sdss 0105@xmath180033 comes from the [ oii ] emission in its spectrum and our adopted redshift for sdss 2204 + 0031 comes from cross - correlation with sdss 0105@xmath180033 .
the only remaining constraint for both objects is that @xmath169 since we do not see ly@xmath22 or the ly@xmath22 forest .
sdss 2215@xmath180045 ( [ sdss2215 ] ) has a nearly unique spectrum , in which feiii absorption is much stronger than feii absorption .
there are several other possibly similar objects with lower snr ( [ morefe3 ] ) , but until better data on them are available we discuss only sdss 2215@xmath180045 in detail .
a full photoionization calculation of the conditions in the bal gas in this object is beyond the scope of the present work , so we merely speculate on what conditions might be needed to produce such a strange spectrum . in sdss 2215@xmath180045 ,
feiii absorption is present from terms with different excitation potentials ( ep ) , and absorption in feiiiuv48 ( ep 5.08ev ) is as strong or stronger than in feiiiuv34 ( ep 3.73ev ) or in feiiiuv50 ( ep 7.86ev ) . using line strengths from & pradhan ( 1996 )
, it is easy to show that this can never happen in conditions of local thermodynamic equilibrium ( lte ) . nor
can it happen if the particle densities are too low for lte to apply , since then the population of the lower term of feiiiuv48 would be reduced even further .
the lower term of the feiiiuv48 multiplet must be overpopulated in sdss 2215@xmath180045 .
this is a rare occurrence ; in most bal quasars where feiii is detected , absorption from feiiiuv34 is stronger than from feiiiuv48 .
for example , in sdss 1723 + 5553 ( figure[f_hiscal ] , top panel ) , uv34 at @xmath01910 is strong but uv48 at @xmath02070 must be much weaker since the 2070 feature can be explained entirely by crii .
sdss [email protected] ( menou et al . 2001 ) also shows feiiiuv34 but not uv48 . according to ( 1996 ) , there are no resonances in the collision strengths to the lower terms of these multiplets strong enough to explain why the lower term of the uv48 multiplet ( @xmath170 ) is overpopulated relative to that of the uv34 multiplet ( @xmath171 ) and the uv50 multiplet ( @xmath172 ) .
thus we conclude that in this object the population of the lower term of the feiiiuv48 multiplet is governed by a fluorescence resonance due to absorption of an emission line ( or possibly by recombination from feiv to excited feiii ) . in either case
the lower term of uv48 could be populated directly or indirectly .
the latter would be via population of an even more excited term followed by a cascade of emission down to the lower term of uv48 , though the lack of uv50 absorption means that any cascade does not involve the lower term of that multiplet .
( 2000 ) show that there is a direct resonance with ly@xmath22 from the ground term of feiii ( @xmath173 ) to the lower term of feiiiuv34 , but that the transitions from the ground term to the lower term of feiiiuv48 are weak and forbidden .
thus , if a fluorescence resonance is at work , it is probably a indirect one .
the other unusual aspect of the absorption in this quasar is that feiii is much stronger than feii , and probably stronger than mgii as well . in most bal regions , where the ionization parameter is large and
the densities may not be high enough for lte to apply , the ratio feiii / feii ( as well as mgiii / mgii and aliii / alii ) will be determined by photoionization equilibrium rather than by the saha equation .
thus , without detailed modeling , we can only conclude that the gas producing the observed bal trough in sdss 2215@xmath180045 must be of moderately high ionization , and with a high enough column density to produce appreciable feiii absorption .
given this requirement , and the observed strength of its absorption relative to feii , feiii is probably the dominant fe ion .
note that such a highly ionized bal region should also have a high temperature ; if so , the observed reddening must be produced by dust outside the bal region . in the & chiang ( 1998 ) and ( 2000 ) disk wind models for bal quasars , detached troughs such as those in sdss 2215@xmath180045
are seen along lines of sight which skim the upper surface of the outflow , which is more highly ionized than the rest of the flow .
the unique highly ionized hibal sbs 1542 + 541 ( telfer et al . 1998 ) might be seen along such a line of sight .
similarly , the uniqueness of sdss 2215@xmath180045 might stem from its being unusual in three ways : a felobal quasar seen along a line of sight through the highly ionized upper surface of a disk wind , and with a column density through the wind high enough so that considerable feiii absorption is seen . alternatively , since at least some quasars experience a phase of high - covering - factor lobal absorption in their youth ( canalizo & stockton 2001 ) , sdss 2215@xmath180045 might be a ` teenage ' quasar captured in the act of ionizing away its youthful bal cocoon .
lastly , the unusual absorption might be due to an unusual spectral energy distribution for this quasar .
photoionization modeling of the bal gas in sdss 2215@xmath180045 and other strong - feiii - absorption bal quasars , especially ones with narrower troughs , is needed to understand what is special about such objects , and why they are so rare .
we are aware of only three bal quasars with extinctions greater than those seen in our extremely reddened bal quasars ( [ extred]).@[email protected] , but it is not known if this object has uv absorption substantial enough to qualify as a bal quasar ( reynolds et al .
hawaii 167 has @[email protected] ( egami et al .
1996 ) and first 0738 + 2750 has @[email protected] ( gregg et al .
2002 ) . first 1556 + 3517 has @[email protected] along the direct line of sight to its central engine , but its observed uv flux is dominated by a scattered component with @[email protected] ( najita , dey , & brotherton 2000 ) .
its uv spectrum therefore does not resemble the extremely reddened bal quasars of [ extred ] ; in fact , it is very similar to the overlapping - trough bal quasars of [ abrupt ] .
the detection of numerous heavily reddened quasars in the sdss lends weight to the growing consensus that a population of red quasars does exist ( e.g. , gregg et al . 2002 ) .
quasars do not have to have bal troughs to be reddened ( richards et al .
2001 ) , but bal quasars are more often reddened than normal qsos are ( richards 2001 ) , and there is evidence for dust in the absorbing region of at least one bal quasar ( de kool et al .
2001 ) . in this section
we note a few constraints our reddened bal quasars ( [ reddened ] ) place on reddening of quasars by dust .
dust reddening does not affect equivalent widths , yet many of these objects lack broad emission lines .
this may indicate that the broad line region is more heavily reddened than the continuum , which could happen if there is a range of extinctions among the sightlines to the emission regions ( hines & wills 1995 ) .
alternatively , the majority of the observed flux could be scattered continuum light , a hypothesis which can easily be tested via polarization measurements .
the five extremely reddened bal quasars we have presented ( [ extred ] ) all have @xmath174 .
similar objects at @xmath175 will usually be too faint for the sdss to detect or target , but it should be able to find lower - redshift analogs .
however , to date we have found no reddened @xmath176 bal quasars which have emission lines as weak as these objects .
simulating such objects colors would help determine if this absence is real or due to their being overlooked by the quasar target selection algorithm .
we have seen evidence in several reddened bal quasars for spectral breaks which might require an extinction curve steeper than that of the smc .
the spectra of hot stars on the spectroscopic plates on which these objects were discovered are unremarkable , so the spectral breaks are not reduction artifacts .
we now examine alternative explanations for the spectral breaks as well as possible origins for such extinction curves .
we have not considered the contaminating effects of host galaxy emission , but it could be important only for objects which are very low luminosity .
sdss 0342 + 0045 is the lowest luminosity object , but the presence of feii absorption indicates it is a bal quasar .
the troughs in it and the other objects are also very deep .
this limits the contribution of host galaxy light to typically @xmath3820% , even assuming that the troughs are resolved and host galaxy light is entirely responsible for the remaining emission in the troughs .
the apparently different reddenings shortward and longward of 2000 in the quasar rest frame are not more easily explained by extinction in a galaxy between us and the quasar instead of by extinction at the quasar redshift . in the ultraviolet through the optical , plausible extinction curves have greater extinctions at shorter wavelengths ( except near the 2200 bump ,
for which there is no evidence in the spectra ) , and the _ slope _ of the extinction curve is also greater at shorter wavelengths .
thus the intervening reddening needed to match the composite spectrum at @xmath212000 would be larger than the intrinsic reddening needed , and even then , the discrepancy at @xmath192000 would be worse .
bal quasars may be preferentially viewed edge - on ( e.g. , glenn , schmidt , & foltz 1994 ) .
limb darkening and relativistic effects should make edge - on accretion disk spectra look different ( czerny & elvis 1987 ; laor & netzer 1989 ; sun & malkan 1989 ; hubeny et al .
however , in all these models , at @xmath55@xmath45912 the dominant effect is limb darkening with a smooth wavelength dependence which does not change the spectra appreciably . only for one choice of parameters considered by & elvis ( 1987 , model 1.5 in their figure 4 )
does a change in viewing angle produce a break in the spectrum similar to those observed here . thus viewing angle effects
are unlikely to be the explanation for our spectra .
( 1994 ) show that pg 0043 + 039 appears to have a continuum break at 2400 ( their figure 4 ) , but is actually a reddened extreme feii emitter ( their figure 5 ) . among our objects
, however , only sdss 1453 + 0029 shows the _ extreme _ feii emission required to mimic such a continuum break .
the objects might be intrinsically much bluer than the typical quasar and even more heavily reddened than we estimate , as small differences in the apparent slopes in different spectral regions are exaggerated by very large reddenings .
however , the distribution of spectral slopes of unreddened quasars limits the reddening which can reasonably be inferred .
for example , if sdss 0318@xmath180600 is reddened by @[email protected] , it would have an intrinsic @xmath177 , bluer than essentially all known quasars .
most of these objects do not match the composites exactly even when dereddened .
this is probably just because quasar continuum and emission line properties vary from object to object .
however , if strong luminosity or redshift dependencies in quasar spectra exist ( e.g. , green , forster , & kuraszkiewicz 2001b ) then the sdss composite quasar spectrum might be an incorrect representation of quasars at @xmath178 because the objects that contribute to the composite spectrum at @xmath192000 are of higher average redshift and luminosity than those that contribute at 2000 - 3000 . any such dependencies would have to be improbably strong to bias the sdss composite significantly , as the sample of quasars used to construct the template has a mean @xmath179 at @xmath180 and @xmath181 at @xmath182 , with only a small number of objects contributing at @xmath183 .
it is also conceivable that the apparent spectral break in the dereddened spectra of these objects is an intrinsic property , rather than an artifact of dereddening with an inappropriate extinction curve .
however , we can think of only two ways to produce such a break , both of which are rather _ ad hoc_. an accretion disk with a gap at radii where @xmath55@xmath192000 emission is produced ( cf . [ gradother ] ) might produce a continuum shape with the broad peak required ( @xmath184 ) .
such a gap would be unstable , and so this idea can be ruled out if no variability is seen .
a spectral break could also conceivably arise if the continuum source has a spatial temperature gradient and its hotter regions are seen through more dust than its cooler regions .
an accretion disk viewed nearly edge on through a dusty coplanar torus is one example . if the inclination angle is such that the line of sight to the black hole has @xmath185 , and thus lines of sight to the near side @xmath186 , then the emission from the far side of the disk will dominate , with decreasing optical depth for lower temperature regions farther out in the disk .
none of the explanations discussed above seem particularly likely , so we conclude that the extinction curve in these objects are steeper than the smc extinction curve at @xmath192000 .
the smc extinction curve is itself steeper than those of the lmc , the mw , and the starburst extinction curve of ( 1994 ) .
smaller average grain sizes are needed to produce steeper extinction curves , since small particles do not scatter efficiently at long wavelengths .
the theoretical extinction curve of 0.01 grains of ` astronomical silicate ' constructed by & foltz ( 1992 ) from the work of ( 1985 ) is steeper than the smc curve .
spherical si or si / sio@xmath108 grains of @xmath010 radius can also produce steep extinction at 1400@xmath19@xmath55@xmath192000 ( li & draine 2002 ) .
observations of our objects at higher snr should be able to better constrain the shape of the observed extinction curve for comparison to theoretical models .
observationally , ( 2001a ) find that the seyfert 1 ngc 3227 has the same extinction at @xmath187 as in the smc but greater extinction at @xmath188 .
however , the slope of its extinction curve at @xmath188 is very similar to that of the smc , so it can not explain our objects .
the heavily reddened gravitationally lensed quasar mg 0414 + 0534 may also show evidence for a very steep uv extinction curve ( lawrence et al .
1995 ; falco et al .
however , ( 1999 ) find substantial differential reddening between the different components of mg 0414 + 0534 , which makes very uncertain any conclusions about extinction curves reached from study of its combined spectrum .
our result is not necessarily in conflict with ( 2001b ) , who found evidence for flat extinction curves in the circumnuclear regions of seyferts , which , marconi , & oliva ( 2001a ) interpret as being produced by dust dominated by very large grains .
the circumnuclear environments of quasars span a wide range of conditions , and it is possible that dust might exist in different dominant forms at different locations , especially if dust can actually be created in quasar outflows ( elvis , marengo , & karovska 2002 ) . on a related note ,
the presence of substantial reddening with detected narrow mgii but not feii absorption , as seen in sdss 1453 + 0029 and sdss 0127 + 0114 ( figure [ f_w1dered ] ) , is anomalous compared to the smc ( e.g. , welty et al .
2001 ) . in the smc , for reddenings as large as those inferred here , feii as well as mgii absorption from gas associated with the dust is consistently detected .
this anomaly is also seen in sdss 2204 + 0031 and sdss 0105@xmath180033 ( figure [ gradered ] ) .
it may be that the dust causing the reddening and the gas causing the absorption lines are not located in the same region .
but if they are , we may be seeing gas which is ionized such that feii and mgii are not the dominant ions of fe and mg ( and so only the strong mgii absorption line is seen ) but which is not so hot or dense that all the dust has not been destroyed ( cf .
sdss 2215@xmath180045 , [ impfe3 ] ) . this idea can be tested with photoionization modeling and by looking for narrow aliii , civ or ovi absorption . to determine how common very steep uv extinction curves are around quasars will require large samples .
for example , using photometry of 104 reddened sdss quasars , not just bal quasars , ( 2001 ) estimate a typical uv extinction curve at least as steep as that of the smc ( prevot et al .
1984 ) . despite their impressively absorbed spectra , overlapping - trough bal quasars ( [ abrupt ] )
do not necessarily have absorbing gas with column densities much larger than other felobals , just gas with a larger velocity range .
the low - ionization gas must have optical depth @xmath186 over a velocity range of at least 5000kms@xmath16 , and possibly @xmath2119000kms@xmath16 , with typical covering factor @xmath090% .
whether or not a given model can produce absorption meeting these requirements is a question for detailed modeling beyond the scope of this paper .
how unusual is the variable absorption in sdss 0437@xmath180045 ( [ weird2 ] ) ?
the only systematic work on variability in bal trough strengths is that of ( 1994 ) , who monitored 23 bal quasars on rest frame timescales of one to several years.057 ( smith & penston 1988 ) , q 1303 + 308 ( vilkoviskij & irwin 2001 ) and pg 0946 + 301 ( arav et al .
2001b ) , the mini - bal q 1159 + 0128 ( aldcroft , bechtold , & foltz 1997 ) , the lobals mrk 231 ( boroson et al . 1991 ) and lbqs 0103@xmath182753a ( junkkarinen et al . 2001 ) , and the possible very high ionization bal pg 1115 + 080 ( michalitsianos , oliversen , & nichols 1996 ) . ]
he found that four showed large bal trough changes ( 20 - 40% in normalized intensity ) .
results for three of these have been published : um 232 ( barlow , junkkarinen , & burbidge 1989 ) , h 0846 + 1540 ( barlow , junkkarinen , & burbidge 1992a ) and cso 203 ( barlow et al .
five other objects showed smaller but still significant bal trough changes , and a further six showed marginal changes .
thus the best available estimate is that on a rest frame timescale of 1 - 3 years , 17@xmath189% of bal quasars show trough intensity variations of 20 - 40% and at least 21@xmath190% more show @xmath1920% variations . given our sample of @xmath020 unusual bal quasars , for about half of which we have multiple spectra , finding one strongly variable object ( up to @xmath015% in two months ) is broadly consistent with the results of ( 1994 ) .
however , sdss 0437@xmath180045 is still very unusual because of both the rapidity with which it has varied and the fact that at least three wide regions have varied , not just one trough .
creating even a few lines of sight with the high optical depth , high covering factor absorption extending over the large velocity range seen in these lobals might require a large global covering factor for the bal gas .
such covering factors are expected if these lobals are recently ( re)fueled quasars seen expelling a dense shroud of dust and gas ( [ frac ] ) .
it is unclear if variability in trough amplitude and velocity in such lobals would be stronger and more common than if the outflow arises from a disk wind .
stable disk winds are often invoked to explain the observed lack of velocity variability in bal outflows , but disk winds may not be particularly stable ( e.g. , proga et al .
observational monitoring of all varieties of bal quasars is needed to establish the range of velocity and trough strength variability seen in each .
theoretical modeling of ` quasar breakout ' from a dense shield of dust and gas is needed to determine whether that model can reproduce the detailed properties of lobal troughs ( variability , partial covering , velocity structure , etc . ) and lobals in general ( e.g. , strong feii and weak [ oiii ] emission ) .
the work of ( 1999 ) is an important start in this respect .
the two low - redshift bal quasars with many narrow troughs ( sdss 1125 + 0029 and sdss 1128 + 0113 ; [ scalloped ] ) were identified as unusual because of those troughs , but the most interesting thing about them is that they seem to show mgii absorption longward of the systemic redshift .
sdss 1125 + 0029 has mgii absorption ranging from 800kms@xmath16 shortward of the host galaxy absorption - line redshift to @xmath181400kms@xmath16 longward , while sdss 1128 + 0113 has mgii absorption ranging from 1400kms@xmath16 shortward to @xmath181700kms@xmath16 longward of the narrow - line redshift ( all velocities are accurate to @xmath37100kms@xmath16 ) . in both objects
the absorption is deepest near the high - velocity ends of the troughs , which is the opposite of what is usually seen for low - ionization lines ( voit et al .
the feii trough shapes in both these objects look similar to the mgii trough shapes ( best seen at the red end of the uv1 multiplet near 2630 ) . for sdss 1128 + 0113 ,
it is remotely possible that the longward - of - systemic absorption could be spurious , but only if the narrow [ oii ] emission line is blueshifted by @xmath01400kms@xmath16 from the true systemic redshift . in nearby agn ,
the narrow forbidden line redshifts are within @xmath37100kms@xmath16 of the host galaxy stellar absorption and hi21 cm emission redshifts ( see discussions in & fan 1992 and 1999 ) .
however , outflow velocities of up to @xmath01000kms@xmath16 have been seen in the narrow - line regions in the central @xmath01kpc around even low - luminosity agn ( veilleux et al .
1994 ) , so more luminous objects could in principle have somewhat higher outflow velocities in their narrow line regions .
a combination of disk obscuration and patchy line - emitting gas in the outflow could then explain why only blueshifted line emission is seen ( e.g. , ngc 3079 ; veilleux , shopbell , & miller 2001 ; cecil et al .
however , for sdss 1125 + 0029 we know that the [ oii ] is blueshifted from the host galaxy absorption lines by only 310@xmath37130kms@xmath16 .
thus blueshifted narrow lines can not explain the longward - of - systemic absorption in that object .
could the longward - of - systemic troughs be due to feii absorption
? this could be the case if feii transitions exist between 2820 to 2800 . to test this possibility
, we have used the feii line lists of ( 1999 ) to construct a toy model of expected feii absorption at the systemic redshift .
we use only the einstein a coefficients to estimate the strength of each line , thus assuming that all levels are equally populated . in reality
, the higher levels which might give rise to absorption near mgii will be less populated than we assume , and thus the strength of such absorption lower .
we created a toy absorption spectrum equal to exp(@xmath191 ) , where @xmath24 is the arbitrarily normalized line strength ` spectrum ' , smoothed by the spectral resolution .
figure[f_scallopedcomp ] compares this theoretical spectrum to the two objects in question .
there is no strong narrow feii trough near 2820 which could explain the apparent longward - of - systemic mgii absorption . blueshifted absorption from feiiuv195,196 at 2855 can not explain it either : there is no corresponding blueshifted feii@xmath552750 absorption at @xmath02695@xmath182715 in sdss 1125 + 0029 , and no corresponding strong blueshifted mgii absorption at @xmath02745@xmath182765 in sdss 1128 + 0113 . the only previously recognized case of a possible longward - of - systemic bal trough , un j1053@xmath180058 ( brotherton et al .
1998a ) , has civ , aliii and mgii troughs extending @xmath01500kms@xmath16 longward of the ciii ] redshift .
however , this redshift could be biased because ciii]@xmath551908 can be blended with emission from aliii@xmath621854,1862 , siiii]@xmath551892 and feiiiuv34,61,68 .
it is also possible that ciii ] is shifted shortward of the systemic redshift in this object : two of eighteen quasars studied by ( 1999 ) have ciii ] shifted @xmath211000kms@xmath16 shortward of [ oiii ] .
a literature search uncovered one previously unrecognized possible longward - of - systemic trough : the mini - bal in 3c 288.1 ( hamann , netzer , & shields 2000 ) , which lies within @xmath371000kms@xmath16 of the broad emission line @xmath192 .
a narrow - line @xmath75 is clearly needed for both these objects .
an indisputable host galaxy redshift for sdss 1128 + 0113 and detailed feii absorption line modeling for it and sdss 1125 + 0029 are needed to firmly rule out other possibilities , but the simplest explanation seems to be that the bal troughs in these objects extend longward of the systemic host galaxy redshifts . infalling gas crossing our line of sight would produce absorption longward of the systemic redshift . if the gas is in individual clouds , high - resolution spectra should show that the apparently smooth troughs around the systemic redshifts in these objects break up into narrow absorption lines .
if the gas is not in individual clouds , then both a low - ionization inflow and a low - ionization outflow are present
. the inflow must be optically thick and have the same maximum local covering factor as the outflow , since the absorption at @xmath193 joins up smoothly to that at @xmath194 both in mgii and in feii lines with much lower optical depths .
( 1999 ) show that inflows and outflows with velocities and possibly column densities comparable to those observed here can coexist in the central regions of agn .
the gas in the flows seen in their isothermal models is too highly ionized to produce the absorption seen here , but it remains to be seen if more refined models explicitly including the thermal balance of the gas can reproduce the observations .
the longward troughs might be due to a freely expanding outflow .
the problem with this is that the maximum trough velocities shortward and longward of the systemic @xmath75 are roughly equal .
the free expansion velocity is therefore much larger than the outflow velocity , leaving us with no explanation for what has accelerated the gas to such high velocities .
a similar problem arises in the model of & norman ( 1995 ) . in
that model bals arise in debris trails from mass - losing stars in the nuclear star cluster , but the observed redshifted velocities in these objects are larger than the central velocity dispersions of any galaxy .
we might instead be viewing an outflow that arises from a distribution of clouds with a substantial virial velocity dispersion . in this picture
, the extent of the absorption longward of the systemic redshift would be the same for troughs of different ionization , but the absorption shortward would depend on the ionization present at different velocities in the outflow , and so could be different for different ions .
( recall that high - ionization absorption typically extends to higher terminal velocities than low - ionization absorption . ) to keep the velocity dispersion @xmath195 from being too large in this case , the clouds must be located far away from the quasar . assuming a @xmath196 black hole and an isotropic velocity field , a distance of @xmath197 cm ( 7 light - months ) is required to match the observed @xmath198kms@xmath16 .
an isotropic velocity dispersion for the cloud population seems most likely : the clouds can not all be on plunging orbits as they would soon be destroyed after approaching too close to the central engine , and it is unlikely that they have a disklike distribution given the observed radial velocity dispersion .
an isotropic velocity dispersion implies that the clouds have a global covering factor of essentially unity , which has been predicted for at least some lobals from other considerations ( canalizo & stockton 2001 )
. however , longward - of - systemic absorption should not be rare in this model .
its apparent rarity must be attributed to the difficulty of obtaining narrow - line redshifts given the weakness of [ oiii ] in lobals ( boroson & meyers 1992 ) .
it is also not obvious if this model can explain why bal troughs do not show variability in velocity as well as amplitude , or if it can explain the similar trough shapes in our two objects , which the next model we discuss might .
a possibly more promising explanation arises from geometry and rotation .
a wind arising from a rotating accretion disk will share the rotation velocity of the disk , and in fact the tangential velocity may initially dominate the vertical and radial velocities . if we have a grazing line of sight to such an accretion disk which passes through a section of the wind where the velocity is predominantly rotational , and if the continuum source is no more than an order of magnitude smaller than the inner radius of the wind , the wind can have both approaching _ and _ receding velocities along our line of sight ( cf .
( proga et al . 2002 ) , where this possibility is mentioned for disk winds in cataclysmic variables ) .
this is demonstrated in figure [ f_view]a , which shows a top view of such a disk .
the inner edge of the wind is at radius @xmath199 , and @xmath200 is the angular coordinate along the surface of the disk measured from our line of sight .
the continuum at some absorption wavelength of interest is assumed to arise predominantly from radii @xmath201 , since the disk beyond @xmath202 will be too cool to contribute significant emission at that wavelength . a wind element at ( @xmath203 )
will have velocity @xmath204 and will absorb continuum emission from a region of length @xmath23 at that velocity .
as seen in the figure , @xmath204 can be redshifted if the rotational velocity dominates .
figure [ f_view]b is a side view of one quadrant of the same disk .
the disk has inclination angle @xmath12 measured from the line of sight toward the disk normal ; i.e. , @xmath205 for an edge - on disk and @xmath206 in figure [ f_view]b .
the wind opening angle is @xmath55 , measured in the @xmath207 direction from the plane of the disk .
the disk , which is optically thick , is shown as the heavy line running diagonally across the entire plot .
the wind arises at radius @xmath199 ( and beyond ) , and fills at most the region outlined by the three thick lines .
the wind may initially be dominated by its vertical velocity @xmath208 , but at the terminal opening angle @xmath55 the radial velocity @xmath209 and the wind has essentially no vertical velocity component .
the wind can absorb flux from a continuum emitting region of size @xmath202 only if the wind rises far enough above the disk to be seen in projection against this region before becoming purely radial .
geometrically , @xmath12 must be large enough to satisfy the inequality @xmath210 which simplifies to @xmath211 this constraint is a weak one which will be satisfied for any line of sight to the black hole which passes through the bal wind .
however , if the continuum source is extended enough , this constraint can also be satisfied for some inclination angles @xmath212 where the line of sight to the black hole itself does not pass through the wind .
( this complicates the conversion from wind covering factor to bal covering factor , but provides an additional way to achieve partial covering of the continuum source . ) for disks with inclination angles @xmath12 large enough to satisfy equation [ e_cosi ] , the wind will be seen in absorption over a range of velocities given by @xmath213\times[\cos(i),\sin(i+\lambda)]\times \cos(\phi)\ ] ] for @xmath214 . here
@xmath215 is the rotational velocity at @xmath199 , @xmath216 is the maximum tangential angle @xmath200 for which the wind shadows the continuum emission region ( @xmath217 ) , and @xmath208 refers only to the initial vertical velocity of the wind and not the component of the radial velocity @xmath218 perpendicular to the disk .
the terms in brackets indicate possible ranges .
acceleration will produce a range of radial velocities @xmath219 as wind elements move from @xmath220 to @xmath221 , with @xmath222 for an initially vertical wind .
geometry produces a range of line of sight projection factors for those radial velocities , from cos(@xmath12 ) at @xmath220 to sin(@xmath12@xmath110@xmath55 ) at @xmath221 , passing through a maximum of unity if @xmath223 .
the @xmath215 term in equation[e_vlos ] is the only one which can be negative .
this equation demonstrates that longward - of - systemic absorption can occur for lines of sight where the wind velocity is dominantly rotational .
the maximum longward - of - systemic velocity will occur if our line of sight passes through bal gas which has not yet experienced significant radial acceleration .
the quasar models of murray _ etal . _ ( & chiang 1998 , and references therein ) and ( 2000 ) both incorporate disk winds , and so longward - of - systemic absorption is possible in both scenarios . for both these models
, we can estimate the maximum possible redshifted velocity and the maximum angle over which longward - of - systemic absorption might be seen by assuming a purely vertical flow with @xmath224 .
we also assume keplerian rotation ( @xmath225 ) and that the 2800 continuum arises from the accretion disk at radii out to @xmath226 cm ( pp .
37 and 45 of peterson 1997 ) .
this is a conservative assumption because we use the radius of peak emission at 2800 as @xmath202 , rather than the largest radius at which significant 2800 emission is produced , and so we slightly underestimate @xmath227 . in the elvis model ,
the wind begins vertically and slowly becomes radial , forming a funnel with a effective wind opening angle of @xmath55=27 .
( 2000 ) gives @xmath228 cm for ngc 5548 , which has @xmath229 ( peterson & wandel 2000 ) .
we take @xmath230 cm since we are primarily concerned with low - ionization absorption , which in this model arises from radii up to twice as large as that of the high - ionization absorption ( figure 5 of 2000 ) .
thus @xmath215=8800kms@xmath16 and @xmath231 .
equation[e_cosi ] then requires @xmath232 for shadowing to occur ; however , the wind is only visible for @xmath233 since the radial part of the flow is optically thick . using @xmath208=6000kms@xmath16 ( 3.2 of 2000 ) ,
equation[e_vlos ] gives @xmath234sin(@xmath12)+6000cos(@xmath12)kms@xmath16 .
this yields possible line of sight velocity ranges from @xmath235kms@xmath16 for @xmath236 to @xmath237kms@xmath16 for @xmath238 ( assuming a disk opening angle of 1 ) .
the maximum negative velocity and the range of velocities for @xmath238 are only a factor of 1.32 higher than the observed velocities in the two sdss quasars with longward - of - systemic absorption . in the murray _
_ model , the wind streamlines rapidly become purely radial , resulting in a wind opening angle of only @xmath55=6 . for @xmath239
the wind begins at @xmath240 cm .
thus , @xmath241 , and shadowing occurs for @xmath242 .
we adopt @xmath208=5000kms@xmath16 since ( 1995 ) state that the wind streamlines are more vertical than radial at such outflow velocities .
equation[e_vlos ] then yields @xmath2431300sin(@xmath12)+5000cos(@xmath12 ) kms@xmath16 . with a disk opening angle of 1 ( murray et al . 1995 ) ,
the ranges of possible velocities range from @xmath244kms@xmath16 for @xmath245 to @xmath246kms@xmath16 for @xmath238 .
the @xmath238 numbers are in good agreement with the observations of the two sdss quasars .
thus , both these models produce estimated ranges of absorption velocities which are qualitatively consistent with those observed in the two sdss quasars with longward - of - systemic absorption .
however , are the @xmath247 values used in those estimates ( @xmath248 ) appropriate for those quasars ? converting the two quasars absolute @xmath249 magnitudes to @xmath250 using @xmath251 , adopting a bolometric correction of @xmath252 magnitudes and assuming a typical quasar value of @xmath253 ( mcleod , rieke , & storrie - lombardi 1999 ) , we estimate @xmath247 values of 4.2 and 3.3 @xmath254 for the two sdss quasars .
this is within a factor of two to three of the values used in the estimates , which is adequate given the uncertainties .
finally , we note that the & begelman ( 1995 ) magnetically confined disk wind model can not explain longward - of - systemic absorption .
their wind arises so far from the continuum source that the source no longer appears appreciably extended . in that model @xmath255 cm
, so @xmath256 and the potentially negative first term in equation [ e_vlos ] is negligible .
we have shown that incorporating an extended continuum source in either the murray _
etal . _ or elvis disk wind models appears capable of explaining redshifted bal absorption . extended continuum sources are a generic result of accretion disk models , though such models are not without their problems ( e.g. , pp .
45 - 55 of peterson 1997 ) .
redshifted absorption requires a very nearly edge - on geometry so that the rotation - dominated base of the wind shadows much of the continuum source . nearly edge - on geometries are of course intrinsically rare , and
many such objects will be hidden by dust if accretion disks are roughly coplanar with the torii of gas and dust thought to exist at larger radii ( schmidt & hines 1999 ) .
thus the observed rarity of redshifted absorption troughs is to be expected .
nonetheless , other longward - of - systemic bal troughs may have gone unrecognized due to confusion with mgii emission and to the scarcity of narrow - line host galaxy redshifts for bal quasars , let alone stellar absorption - line redshifts . in particular , some strong @xmath257 associated absorption systems in radio - loud quasars ( anderson et al .
1987 ) could be redshifted mini - bals .
there are several ways of testing our model .
shorter continuum wavelengths are produced at smaller @xmath202 in standard disk models . since a smaller
@xmath202 will decrease the maximum observable redshifted velocity , troughs at shorter wavelengths should have smaller longward extents if standard disk models are correct .
in fact , near - simultaneous uv - optical variability suggests that at least part of the uv - optical continuum emission is reprocessed high - energy emission ( peterson 1997 , p. 48 ) , in which case all troughs might have the same longward extents .
we can not determine which is the case from existing data .
in sdss 1125 + 0029 and sdss 1128 + 0113 the snr and/or resolution are insufficient to study any trough besides mgii , given the huge number of feii and other transitions in the wavelength range observed .
high - resolution data for higher - ionization transitions in the two low - redshift objects would also help test the virial velocity dispersion model
. the longward - of - systemic absorption should be the same for all lines in that model , while confirmation in _ both _ objects of o - star - like absorption profiles ( deepest at high outflow velocities ) would be difficult to understand because the absorption profiles in different objects should be uncorrelated since the gas is on random orbits .
more detailed comparisons with the spectra might be possible if some effects we have neglected are considered ( e.g. , disks so thick that the disk opening angle is comparable to or greater than the wind opening angle ) .
elliptical and/or warped disks might be needed to explain why the observed mgii troughs extend farther longward of systemic than shortward ; in the model presented herein any such asymmetry should be to the blue . also , the continuum emission region may be more of an annulus than a disk , since the inner disk will be very hot and may not contribute significant emission at the wavelength in question . for a narrow annulus , the path length @xmath23 shadowed by the bal flow may be smaller at @xmath258 than near the maxima @xmath259 .
similarly , the column density through the bal flow will have a weak minimum at @xmath258 if its gas density is constant with @xmath200 .
this may explain the local minimum in the absorption at 2800 in sdss 1128 + 0113 . finally , limb darkening ( hubeny et al . 2000 ) and other radiative transfer effects ( e.g. , murray & chiang 1997 ) will affect the disk emission source function @xmath260 and thus the amount of emission available for absorption by a wind element at ( @xmath203 ) .
for example , relativistic beaming will brighten the approaching side of the disk and dim the receding side ( e.g. , hubeny et al . 2000 ) .
if this effect is the dominant one and the bal region is optically thick , the absorption profiles will be deeper at shorter wavelengths .
such profiles are seen in both low - redshift objects and possibly in sdss 1723 + 5553 , though in that case the shape may be due to unabsorbed broad mgii emission ( arav et al .
nonetheless , if the bal regions in these objects are probing different regions of the quasar accretion disk as a function of velocity , then they have the potential to provide powerful constraints on accretion disk models .
having selected different types of unusual bal quasars from early sdss data , we can ask what fraction of the quasar population they form . in the redshift range @xmath261 where selection of both hibals ( via civ ) and lobals ( via mgii or aliii ) is possible ,
there are 1807 quasars in the sdss edr quasar sample ( schneider et al .
2002 ) . in this
redshift range our sample of bal quasars ( reichard et al .
2002 ) includes 284 - 306 bal quasars ( 15.7%-16.9% ) by the ai definition , or 181 - 219 ( 10.0%-12.1% ) by the bi definition .
the lower numbers are firm figures , while the higher numbers include borderline cases ( low snr , preliminary indices less than required by only 1@xmath195 , etc . ) for comparison , at somewhat lower average redshifts the lbqs survey finds a bal quasar fraction of 8 - 11% ( weymann 1997 ) and the fbqs survey 14 - 18% ( becker et al . 2000 ) . the number of lobals at @xmath261 in the sdss edr is 28 by the ai definition or 24.5 ( the 0.5 is a borderline case ) by the bi definition .
thus lobals make up [email protected]% of the sdss edr sample at @xmath261 , in agreement with canonical numbers ( [ intro ] ) .
there are four felobals in the edr in this redshift range , and three or four more at lower @xmath75 .
there are two unusual felobals in the edr in this redshift range : sdss 1723 + 5553 ( [ scalloped ] ) and sdss 1730 + 5850 ( [ abrupt ] ) , and two or three more at lower @xmath75 : sdss 1125 + 0029 ( [ scalloped ] ) , sdss 0300 + 0048 ( [ abrupt ] ) , and possibly sdss 0105@xmath180033 ( [ gradual ] ) .
the other unusual bal quasars in the edr are all at @xmath262 : sdss 1214@xmath180001 ( [ fe3 ] ) , sdss 1453 + 0029 and sdss 0127 + 0114 ( [ weird1 ] ) .
thus , preliminary numbers from the ( incomplete ) sdss edr quasar sample indicate that unusual lobal quasars are about equal in number to felobals and that 50% of all felobals fall into one of our categories of unusual objects . since most of these unusual bal quasars are considerably redder than the targets of most quasar surveys , and since felobals comprise @xmath015% of lobals , most previous samples of lobal quasars are likely to be incomplete by @xmath4515% .
this incompleteness would have little effect on the overall numbers of quasars in any survey , since even a factor of two increase in the number of lobals would translate to only a few percent more quasars overall , but it has probably biased our view of the range of column densities and velocity widths spanned by bal outflows .
note that our numbers are in agreement with & brunzendorf ( 2001 ) , who used an optical variability and zero proper motion survey to limit the fraction of ` unusual ' quasars to @xmath263% for @xmath264 .
we have presented over twenty unusual broad absorption line quasars selected from @xmath3815% of the eventual sloan digital sky survey database .
these objects confirm that several _ populations _ of bal quasars with unusual properties exist , and that the range of parameter space spanned by bal outflows is larger than previously realized .
even so , these objects are all low - ionization bal quasars ( lobals ) .
a corresponding population of hibals with extensive absorption from only high - ionization transitions could exist , but the rest - frame optical - uv region probed by sdss spectra is not very sensitive to such absorption .
the total population of unusual lobal quasars is at least as extensive as that of lobal quasars with absorption from excited - state feii or feiii about 15% of lobals and about half of such felobals qualify as unusual ( [ percent ] ) .
unusual bal quasars are thus not an enormous population , but their properties may provide more stringent tests of bal outflow models than do the more numerous ` normal ' bal quasars .
the unusual bal quasars we have found to date can be divided into four or five categories .
+ @xmath73 two objects with many narrow troughs show mgii absorption extending longward of their systemic host galaxy redshifts by @xmath01200kms@xmath16 .
we favor an explanation where this is due to absorption of an extended continuum source by the rotation - dominated base of a disk wind , but other explanations are possible ( [ impscal ] ) .
+ @xmath73 five objects have absorption which removes an unprecedented @xmath090% of all flux shortward of mgii ( [ abrupt ] ) .
the absorption in one of them has varied across the ultraviolet with an amplitude and rate of change among the greatest ever seen in bal quasars ( [ impabru ] ) .
this same object may also show broad h@xmath1 absorption ( [ weird2 ] ) .
these objects may not be fundamentally new , but they do show that low - ionization absorption can have outflow velocities as large as any ever seen in high - ionization lines .
+ @xmath73 numerous heavily reddened
bal quasars have been found , including two reddened mini - bals with very strong feii emission ( [ weird1 ] ) .
the five reddest objects are reddened by @xmath2@xmath3 ( [ extred ] ) , and in two of them we find strong evidence that the extinction curve is even steeper than that of the smc ( [ extcurvered ] ) .
+ @xmath73 we have found at least one object with absorption from feiii but not feii , which may be due to an unusually high column density of moderately high - ionization bal gas ( [ fe3 ] ) . also , the relative strengths of excited term feiii absorption in these objects is different than in most bals with feiii absorption , and can not be achieved in lte , so some sort of resonance must be at work ( [ impfe3 ] ) .
lastly , we have found two luminous , probably reddened high - redshift objects which may be bal quasars whose troughs partially cover different regions of the continuum source as a function of velocity ( [ gradexp ] ) .
many of these objects show absorption from neutral atoms such as mgi and hei , which can be used to constrain the density and ionization in the bal gas . sometimes the distance of the bal gas from the central ionizing source can also be constrained , assuming photoionization equilibrium with that source ( arav et al .
we have presented three new bal quasars with hei absorption ( table [ t_info ] ) , and possibly two more ( sdss 1730 + 5850 and sdss 0318@xmath180600 ) . also , all five of our overlapping - trough bal quasars ( including sdss 1730 + 5850 ) have mgi absorption , and sdss 0318@xmath180600 may have it as well .
this substantially increases the number of objects known to exhibit absorption in these valuable transitions.2852 and hei absorption : the seyfert1 / lobal mrk 231 ( rudy , stocke , & foltz 1985 ) and the felobals q 2359@xmath181241 ( arav et al .
2001a ) , fbqs 1044 + 3656 ( de kool et al .
2001 ) and pss 1537 + 1227 ( djorgovski et al .
2001 ) . the only other previously known agn with hei absorption are the seyfert1 ngc 4151 ( anderson 1974 ) and perhaps the reddened radio - loud quasar 3cr 68.1 ( brotherton et al .
1998b ) . the other previously known agn with mgi absorption are the felobals first 1556 + 3517 ( becker et al .
1997 ) , fbqs 1408 + 3054 ( white et al .
2000 ; becker et al .
2000 ) and possibly tol 1037@xmath182703 ( srianand & petitjean 2001 ) , plus the mini - bals arp 102b ( halpern et al . 1996 ) , 3c 191 ( hamann et al . 2001 ) , and possibly fbqs 1427 + 2709 and fbqs 1055 + 3124 ( white et al .
2000 ) . ]
we close with a mention of some potentially fruitful avenues of research on these unusual lobal quasars .
near - ir spectroscopy and spectropolarimetry of all these objects would help determine if they fit into a geometrical ` dusty torus ' picture of bal quasars ( schmidt & hines 1999 ) or whether they represent an early ` cocoon ' phase in the evolution of individual quasars ( becker et al . 2000 ) .
high resolution spectroscopy of the longward - of - systemic absorption troughs in sdss 1125 + 0029 and sdss 1128 + 0113 could differentiate between several models for bal outflows . despite the broad and blended absorption in overlapping - trough bal quasars , targeted high resolution spectroscopy of caii in sdss 0300 + 0048 and hei in sdss 1730 + 5850 might determine if the column densities as well as the outflow velocities are very large in these objects .
photoionization modeling including detailed treatment of feii and feiii may be needed to understand feiii - dominant bal quasars , particularly the resonance at work in at least sdss 2215@xmath180045 .
lastly , detailed modeling of disk winds and turbulent outflows is needed to determine if such models can explain the wide range of properties seen in these objects , and where such models need to be revised if they can not .
the sloan digital sky survey ( sdss ) is a joint project of the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the max - planck - institute for astronomy , the max - planck - institute for astrophysics , new mexico state university , princeton university , the united states naval observatory , and the university of washington .
apache point observatory , site of the sdss telescopes , is operated by the astrophysical research consortium .
funding for the project has been provided by the alfred p. sloan foundation , the sdss member institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. we thank an anonymous referee , f. barrientos , m. brotherton , c. carilli , d. clements , b. draine , d. morton , m. rupen , j. veliz , k. verner , m. vestergaard , b. wills and h. yee for discussions and assistance .
this research has made use of nasa s astrophysics data system , the atomic line list v2.04 at http://www.pa.uky.edu/$\sim$peter/atomic/ , and the astronomical data center at nasa goddard space flight center .
the apache point observatory 3.5-meter telescope is owned and operated by the astrophysical research consortium .
the w. m. keck observatory is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration , made possible by the generous financial support of the w. m. keck foundataion .
ukirt is operated by the joint astronomy centre on behalf of the uk particle physics and astronomy research council .
the cfht is a joint facility of the national research council of canada , the centre national de la recherche scientifique of france and the university of hawaii .
the two micron all sky survey ( 2mass ) is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
the first survey is suppported by grants from the national science foundation ( grant ast-98 - 02791 ) , nato , the national geographic society , sun microsystems , and columbia university .
pbh acknowledges financial support from chilean grant fondecyt/1010981 and a fundacin andes grant , mas from nsf grant ast-0071091 , and dps and gtr from nsf grant 99 - 00703 .
\(1991 ) defined a ` balnicity index ' or bi with the aim of providing a continuous measure of the strength of the broad absorption while excluding intervening and associated absorption systems .
the most physically relevant quantity for gauging bal strength is the column density in the bal system(s ) , but this can be measured only in some objects , and even then requires high - resolution spectroscopy and detailed modeling .
the bi is a useful substitute , being an observable quantity calculable from low - resolution data .
the bi is calculated from civ ( treating it as a single transition at 1549 ) as follows . define the quasar @xmath75 using mgii , ciii ] , or civ ( in order of preference ) , fit a continuum between siiv / oiv ] and civ , and interpolate over non - civ absorption in that wavelength region . between 3000kms@xmath16 and 25000kms@xmath16 shortward of 1549 at the systemic redshift , measure the modified equivalent width ( in kms@xmath16 ) of the portions of contiguous absorption troughs exceeding 2000kms@xmath16 in velocity .
the equivalent width is ` modified ' because ` absorption ' is defined as only those parts of the troughs which dip at least 10% below the adopted continuum level .
thus the maximum bi is 20000kms@xmath16 .
the criterion of a 10% dip for absorption was established to ensure that false bal troughs are not identified by placing the continuum too high ; the 2000kms@xmath16 contiguity condition to avoid strong intervening narrow - line absorption complexes ; the 3000kms@xmath16 red integration limit to avoid strong associated narrow - line civ@xmath621548,1550 complexes ; and the 25000kms@xmath16 blue integration limit to avoid confusion with siiv / oiv]@xmath551400 emission and siiv bal troughs .
it is worth considering exactly what the ` ideal ' balnicity index is that we are trying to measure .
the ideal bi would be found by measuring the standard equivalent width ( in rest - frame kms@xmath16 at the quasar systemic redshift ) of all confirmed civ bal troughs shortward of the systemic redshift .
the ideal bi would have the following advantages over the standard bi : + @xmath73 it would not exclude bal quasars with broad , shallow troughs such as um 660 ( turnshek 1988 ) .
+ @xmath73 it would not exclude bal troughs within 3000kms@xmath16 of the systemic velocity .
+ @xmath73 it would not exclude ` mini - bal ' troughs ( hamann 2000 ) which have contiguous width @xmath192000kms@xmath16 but otherwise share all the characteristics of bal troughs .
+ @xmath73 it can be calculated for high - velocity bal quasars with troughs that extend to @xmath2125000kms@xmath16 but are not confused with troughs from other transitions ( e.g. , pg 2302 + 029 ; 1996 ) , the drawback of the ideal bi , of course , is that it requires exact knowledge of the quasar s systemic velocity and continuum shape , and of which absorption troughs arise in bal outflows as opposed to being intervening systems ( of civ or other species ) or intrinsic systems unrelated to the bal outflow .
\(1991 ) note that ... better resolution data , coupled with a more refined definition of balnicity , are required to treat ... borderline cases . "
since sdss spectra have a wavelength resolution good enough to identify many intervening absorption systems , we have defined a provisional revised balnicity index that makes more optimal use of sdss data . since our goal is to measure the strength of all intrinsic absorption , not just broad intrinsic absorption , we refer to this revised balnicity index as the _ intrinsic absorption index _ or ai .
the final values of some of the parameters used in the ai may change based on the results of ongoing detailed studies of sdss bal quasars , but the general framework for choosing those parameters is presented here .
the specific issues addressed in the definition of the ai are as follows . in some objects
the mgii emission line @xmath75 is uncertain , let alone that of ciii ] or civ .
we adopt the host galaxy stellar absorption - line redshift whenever possible ; otherwise , we follow ( 2001 ) and adopt as systemic the redshift of [ oiii]@xmath555008 ( vacuum @xmath55 ) , or [ oii]@xmath623727,3729 if [ oiii ] is unavailable .
we correct the @xmath75 measured from other lines to this systemic @xmath75 using the velocity shift for that line as measured in the sdss composite quasar ( vanden berk et al .
these velocity shifts may change slightly as more objects are added to this composite , but a greater concern to be addressed when more quasars are in hand is whether different quasar subtypes have different velocity shifts .
we begin with a fifth - order polynomial fit to the windows discussed in ( 1991 ) : 1575 - 1625 , 1800 - 1820 , 1975 - 2000 , 2140 - 2155 , 2190 - 2200 , 2240 - 2255 , 2665 - 2695 plus a window longward of mgii .
the windows do not always sample the continuum , so they are adjusted where necessary to reach an acceptable final fit .
note that in most cases it is not possible to accurately include broad emission lines in the fit , and so the depth of the absorption will be underestimated where the absorption overlaps in wavelength with the broad emission .
it may be possible to automate the continuum determination for even heavily absorbed bal quasars by comparison with weaker bal quasars as well as non - bal quasars , perhaps along the lines of & szalay ( 1999 ) .
it may also be worthwhile to use a 3@xmath195 significance criterion for determining the starting and ending points of absorption troughs , which would also allow the minimum detectable ai for a given spectrum to be defined without requiring an arbitrary 10% dip below the continuum level .
for now , we use the same modified equivalent width as the bi since empirically we find it reduces the identification of random fluctuations as bal troughs .
a 10% systematic uncertainty in continuum placement is reasonable in most cases , so we adopt that as a characteristic systematic error until a larger number of sdss quasar spectra are available to study the issue statistically .
\(2000 ) points out that mini - bal troughs and confirmed intrinsic narrow absorption line ( nal ) troughs are found over a range of ejection velocities comparable to those of bal troughs .
we wish to measure the strength of such intrinsic , outflowing systems while ignoring intervening systems and associated systems from gas in or near the quasar host galaxy and its environs .
confirmation that a given system is intrinsic requires detection of nonblack saturation , time variability on timescales of years , well - resolved smooth profiles which are broad compared to thermal line widths , densities @xmath21100@xmath265 inferred from excited - state absorption , or possibly very high metallicities ( e.g. , hamann et al .
most of these measurements are beyond the capabilities of sdss spectra , but with an instrumental resolution of @xmath266kms@xmath16 we can resolve individual narrow civ and mgii systems ( doublet separations 498kms@xmath16 and 769kms@xmath16 , respectively ) .
thus in many cases we will be able to directly determine whether an absorption feature @xmath192000kms@xmath16 wide is a bal trough or a complex of narrow absorption systems .
for the ai we therefore adopt a 450kms@xmath16 contiguity criterion instead of the 2000kms@xmath16 criterion used for the bi .
we also slightly revise the definition of contiguous to include _ all _ absorption from troughs @xmath267kms@xmath16 wide , rather than only the absorption beyond the first 2000kms@xmath16 .
we feel the odds of finding non - bal systems in such close proximity to bal systems are small enough to make an occasional overestimate of the ai acceptable .
however , we do consider the 450kms@xmath16 value of the contiguity criterion to be preliminary and subject to revision pending more detailed future studies of sdss bal quasars .
the ai can be calculated for any transition , but for consistency with the bi we use civ whenever possible . when another transition is used , it should be noted in parentheses : e.g. , ai(mgii ) . for objects with no civ data ,
siiv is the best alternative for hibals , and mgii or aliii for lobals .
mgii or aliii absorption can also be used for objects where the continuum around civ is very difficult to estimate .
low - ionization troughs such as mgii and aliii are typically narrower than high - ionization troughs , but eventually it may be possible to correct for this effect statistically . after defining the systemic redshift and the continuum level and interpolating over all absorption features besides the relevant trough , the ai and statistical uncertainty are calculated as @xmath268 ~c ' ~dv \\ \sigma_{\rm ai}^2 = \int_{0}^{(25,000 ) } ( \sigma_{f(v)}/0.9)^2 ~c ' ~dv\end{aligned}\ ] ] where @xmath269 and @xmath270 are the normalized flux and uncertainty ( unsmoothed whenever possible ) as a function of velocity in kms@xmath16 from the systemic @xmath75 .
thus , like the bi , the ai has units of kms@xmath16 .
the integral begins at the systemic redshift ( @xmath271 ) at the wavelength of the shortest - wavelength line of any multiplet involved ( e.g. , 1548.20 for civ or 2796.35 for mgii ) .
the integral extends beyond the highest velocity intrinsic civ system , but in practice the maximum velocity will often be limited by confusion with siiv troughs at 25000kms@xmath16 , so we have noted this in parentheses as a typical limit .
the value of @xmath272 is unity in contiguous intervals of width 450kms@xmath16 or greater where the quantity in brackets is everywhere positive ; otherwise @xmath273 .
the ai is therefore linear from 450kms@xmath16 to @xmath2125,000kms@xmath16 , the latter being a lower limit when the civ absorption appears to extend beyond siiv but is confused with the siiv bal troughs .
the ai can also be extended to @xmath19450kms@xmath16 using higher resolution spectra .
finally , the presence of bal troughs longward of the systemic redshift ( [ rotscal ] ) can be indicated with a superscripted + , e.g. , 3000@xmath274 .
if our explanation for this phenomenon is correct ( [ impscal ] ) , the ai in such cases should still be measured from the systemic redshift .
in practice the statistical uncertainty is dwarfed by the systematic uncertainty in the continuum placement .
nonetheless we quote errors for the ai since they are useful for marginal objects , and since detailed study of large bal and non - bal quasar samples may eventually reduce the systematic uncertainty to a level comparable to the statistical uncertainty . iraf and sm code for determining a continuum , normalizing a spectrum by it , and calculating its ai or bi value is available , along with all spectra of unusual bal quasars presented in this paper ( [ select ] ) , from the contributed data section of the sdss archive at http://archive.stsci.edu / sdss/. , s. g. , brunner , r. j. , mahabal , a. a. , odewahn , s. c. , de carvalho , r. r. , gal , r. r. , stolorz , p. , granat , r. , et al . , 2001 , in to appear in mining the sky , eds .
a. banday et al .
( berlin : springer - verlag ) , in press ( astro ph/0012489 ) , p. b. , gunn , j. e. , knapp , g. r. , narayanan , v. k. , strauss , m. a. , anderson , s. f. , vanden berk , d. e. , heckman , t. m. , et al .
, 2001 , to appear in mass outflow in active galactic nuclei : new perspectives , eds .
d. m. crenshaw , s. b. kraemer , and i. m. george , in press ( astro - ph/0107182 ) rlrrlll ly@xmath22 & 1215.67 & 0.0 & 13.6 & hi & common + nv & 1238.82,1242.80 & 77.5 & 97.9 & hi & common + siii & 1260.42 ( uv4 ) & 8.1 & 16.3 & lo & unusual + siii * & 1264.74,1265.00 ( uv4 ) & ( 0.05 ) & 16.3 & lo & rare + oi & 1302.17 & 0.0 & 13.6 & lo & unusual + siii & 1304.37 ( uv3 ) & 8.1 & 16.3 & lo & unusual + siii * & 1309.28 ( uv3 ) & ( 0.05 ) & 16.3 & lo & rare + cii & 1334.53 ( uv1 ) & 11.3 & 24.4 & lo & unusual + cii * & 1335.66,1335.71 ( uv1 ) & ( 0.01 ) & 24.4 & lo & rare + siiv & 1393.76,1402.77 & 33.5 & 45.1 & hi & common + siii & 1526.73 ( uv2 ) & 8.1 & 16.3 & lo & unusual + siii * & 1533.43 ( uv2 ) & ( 0.05 ) & 16.3 & lo & rare + civ & 1548.20,1550.77 & 47.9 & 64.5 & hi & ( defines ) + alii & 1670.79 & 6.0 & 18.8 & lo & unusual + niii ( * ) & 1709.6,1741.5,1751.9,1773.9 ( uv5,4,3 ) & 7.6 & 18.2 & felo ?
& rare + pi & 1774.95,1782.83,1787.65 ( uv1 ) & 0 & 10.5 & felo ? &
rare + siii & 1808.01 ( uv1 ) & 8.1 & 16.3 & lo & unusual + aliii & 1854.72,1862.79 & 18.8 & 28.4 & lo & common + feiii & 1895.46,1914.06,1926.30 ( uv34 ) & ( 3.7 ) & 30.7 & felo & common + mgi & 2026.48 & 0.0 & 7.6 & felo ? & rare + znii & 2026.14,2062.66 & 9.4 & 18.0 & felo ? &
rare + crii & 2056.25,2062.23,2066.16 & 6.8 & 16.5 & felo ?
& rare + feiii & 2062.21,2068.90,2079.65 ( uv48 ) & ( 5.1 ) & 30.7 & felo & unusual + fei & 2167.45 ( uv21 ) & 0.0 & 7.9 & felo ? &
rare + niii & 2225 ( uv12,13 ) & 7.6 & 18.2 & felo ? &
rare + coii & 2300 ( uv9 ) & 7.9 & 17.1 & felo ? &
rare + niii & 2300 ( uv11 ) & 7.6 & 18.2 & felo ? &
rare + mnii & 2576.88,2594.50,2606.46 & 7.4 & 15.6 & felo ?
& rare + crii & 2680 ( uv7,8 ) & ( 1.48 - 1.55 ) & 16.5 & felo ? & rare + mgii & 2796.35,2803.53 & 7.6 & 15.0 & lo & ( defines ) +
mgi & 2852.96 & 0.0 & 7.6 & felo ? &
rare + crii & 2860 ( uv5 ) & ( 1.48 - 1.55 ) & 16.5 & felo ? & rare + hei & 2945.97,3188.67,3889.74 & ( 19.8 ) & 24.6 & lo & rare + oiii & 3133.70 & ( 36.9 ) & 54.9 & lo & rare + caii & 3934.78,3969.59 & 6.1 & 11.9 & lo & rare + nai & 5891.58,5897.56 & 0.0 & 5.1 & lo & rare + + + feii & 1570 ( uv44,45,46 ) & ( 0.15 - 0.25 ) & 16.2 & felo & rare + feii ( * ) & 1608.45 ( uv8 ) & 7.9 & 16.2 & lo & rare + feii & 1710 ( uv38 ) & ( 0.23 - 0.38 ) & 16.2 & felo & rare + feii ] & 1781.70 ( uv67 ) & ( 1.07 ) & 16.2 & felo & rare + feii & 1785.27,1786.75,1787.996 ( uv191 ) & ( 2.88 ) & 16.2 & felo & rare + feii & 2151.8,2153.0,2177.7 ( uv106 ) & ( 2.27 - 2.33 ) & 16.2 & felo & rare + feii & 2164.34,2173.72 ( uv79 ) & ( 1.66 - 1.69 ) & 16.2 & lo & unusual + feii ( * ) & 2249.88 ( uv5 ) & 7.9 & 16.2 & lo & unusual + feii ( * ) & 2260.78 ( uv4 ) & 7.9 & 16.2 & lo & unusual + feii & 2298.93 ( uv133 ) & ( 2.63 ) & 16.2 & felo & rare + feii ( * ) & 2344.21 ( uv3 ) & 7.9 & 16.2 & lo & unusual + feii ( * ) & 2374.46,2382.77 ( uv2 ) & 7.9 & 16.2 & lo & unusual + feii & 2382.90,2388.39 ( uv117 ) & ( 2.51 - 2.57 ) & 16.2 & felo & rare + feii * & 2400,2600 ( uv2,1 ) & ( 0.05 - 0.12 ) & 16.2 & felo & ( defines ) + feii & 2420 ( uv35,36 ) & ( 0.23 - 0.39 ) & 16.2 & felo & rare + feii & 2460 ( uv209 ) & ( 3.14 - 3.22 ) & 16.2 & felo & rare + feii & 2580 ( uv64 ) & ( 0.98 - 1.09 ) & 16.2 & felo & rare + feii ( * ) & 2586.65,2600.17 ( uv1 ) & 7.9 & 16.2 & lo & unusual + feii & 2750 ( uv62,uv63 ) & ( 0.98 - 1.09 ) & 16.2 & felo & common + feii & 2880 ( uv61 ) & ( 0.98 - 1.07 ) & 16.2 & felo & rare + feii & 2950 ( uv60 ) & ( 0.98 - 1.09 ) & 16.2 & felo & rare + feii & 2985 ( uv78 ) & ( 1.66 - 1.72 ) & 16.2 & felo & rare + feii & 3180 ( opt7 ) & ( 1.66 - 1.72 ) & 16.2 & felo & rare + feii & 3200 ( opt6 ) & ( 1.66 - 1.72 ) & 16.2 & felo & rare + lclllccrrcccccc [email protected] & [email protected] & ? & 0 & 0 & y & 11011 & 4.59 & @xmath1826.59 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.034 + [email protected] & [email protected] & red , hei & 0 & 50@xmath371 & y & 01000 & 1.06 & @xmath1826.84 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.027 + [email protected] & [email protected] & feiii ? & 56 & 2930@xmath375 & y & 10000 & @xmath191.05 & @xmath1825.76 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.038 + [email protected] & [email protected] & ot & @xmath2111520 & @xmath2115347@xmath373 & y & 10000 & @xmath190.93 & @xmath1827.23 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.089 + [email protected] & [email protected] & red , fe & 4017 & 6294@xmath376 & n & 10000 & & @xmath1828.15 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.050 + [email protected] & [email protected] & red , fe & 360 & 476@xmath376 ( mgii ) & y & 10000 & & @xmath1826.74 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.104 + [email protected] & [email protected] & red , fe & 0 & 437@xmath374 & n & & & @xmath1824.63 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.123 + [email protected] & [email protected] & ot & 16330 & 20767@xmath373 & n & & @xmath190.071 & @xmath1826.40 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.034 + [email protected] & [email protected] & feiii ? & 5670 & 6699@xmath375 & n & 10000 & @xmath191.45 & @xmath1826.93 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.047 + [email protected] & [email protected] & ot &
@xmath2113050 & @xmath2117151@xmath371 ( mgii ) & n & & @xmath190.95 & @xmath1823.53 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.057 + [email protected] & [email protected] & red & 760 & 4084@xmath3712 & n & 11010 & 2.23 & @xmath1827.01 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.035 + [email protected] & [email protected] & red & 0 & 1630@xmath3780 ( mgii ) & n & 10010 & & @xmath1826.70 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.028 + [email protected] & [email protected] & mnt , hei & 0 & 491@xmath274@xmath371 & y & 10011 & @xmath190.99 & @xmath1825.54 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.031 + [email protected] & [email protected] & mnt & 0 & 696@xmath274@xmath371 & s & 10010 & @xmath190.98 & @xmath1825.30 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.034 + [email protected] & [email protected] & ot & @xmath2113840 & @xmath2113840@xmath3710 ( mgii ) &
n & 10010 & @xmath190.99 & @xmath1826.97 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.027 + [email protected] & [email protected] & feiii & 2880 & 3149@xmath374 ( mgii ) & y & 01011 & 1.79 & @xmath1825.19 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.020 + [email protected] & [email protected] & red & 0 & 1360@xmath37280 ( mgii ) & n & 10000 & @xmath191.01 & @xmath1826.02 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.039 + [email protected] & [email protected] & red , hei & 0 & 253@xmath18477 ( mgii ) & y & 10000 & @xmath190.054 & @xmath1824.95 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.042 + [email protected] & [email protected] & red & 1350 & 4525@xmath3710 & s & 11010 & 8.56 & @xmath1826.21 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.045 + [email protected] & [email protected] & mnt & 4350 & 7120@xmath3720 & y & 10010 & @xmath190.93 & @xmath1826.96 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.035 + [email protected] & [email protected] & ot , hei ?
& @xmath2110900 & @xmath2115500@xmath37100 ( mgii ) & y & 00010 & & @xmath1824.54 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.030 + [email protected] & [email protected] & ? & 0 & 0 & n & 11010 & 2.94 & @xmath1827.83 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.055 + [email protected] & [email protected] & feiii & 1160 & 1293@xmath373 ( mgii ) & n & 10001 & @xmath190.94 & @xmath1828.42 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0.102 + | the sloan digital sky survey has confirmed the existence of populations of broad absorption line ( bal ) quasars with various unusual properties .
we present and discuss twenty - three such objects and consider the implications of their wide range of properties for models of bal outflows and quasars in general .
we have discovered one bal quasar with a record number of absorption lines .
two other similarly complex objects with many narrow troughs show broad mgii absorption extending longward of their systemic host galaxy redshifts .
this can be explained as absorption of an extended continuum source by the rotation - dominated base of a disk wind .
five other objects have absorption which removes an unprecedented @xmath090% of all flux shortward of mgii .
the absorption in one of them has varied across the ultraviolet with an amplitude and rate of change as great as ever seen .
this same object may also show broad h@xmath1 absorption .
numerous reddened bal quasars have been found , including at least one reddened mini - bal quasar with very strong feii emission .
the five reddest objects have continuum reddenings of @xmath2@xmath3 , and in two of them we find strong evidence that the reddening curve is even steeper than that of the smc .
we have found at least one object with absorption from feiii but not feii .
this may be due to a high column density of moderately high - ionization gas , but the feiii level populations must also be affected by some sort of resonance .
finally , we have found two luminous , probably reddened high - redshift objects which may be bal quasars whose troughs partially cover different regions of the continuum source as a function of velocity . |
the set of possible spectra for the sum of two deterministic hermitian matrices @xmath0 and @xmath1 depends in complicated ways on the spectra of @xmath0 and @xmath1 ( see @xcite ) . nevertheless , if one adds some randomness to the eigenspaces of @xmath1 then , as @xmath2 becomes large , free probability provides a good understanding of the behavior of the spectrum of this sum .
more precisely , set @xmath3 , where @xmath4 is a random unitary matrix distributed according to the haar measure on the unitary group @xmath5 , and suppose that the empirical eigenvalue distributions of @xmath0 and @xmath1 converge weakly to compactly supported distributions @xmath6 and @xmath7 , respectively . building on the groundbreaking result of voiculescu @xcite ,
speicher proved in @xcite the almost sure weak convergence of the empirical eigenvalue distribution of @xmath8 to the free additive convolution @xmath9 .
this convolution is again a compactly supported probability measure on @xmath10 .
a similar result holds for products of matrices : if @xmath11 are in addition assumed to be nonnegative definite , then the empirical eigenvalue distribution of @xmath12 converges to the free multiplicative convolution @xmath13 , a compactly supported probability measure on @xmath14 .
( we recall that @xmath12 and @xmath15 have the same eigenvalue distribution , and that @xmath16 is a commutative operation . ) finally , if @xmath17 and @xmath18 are deterministic unitary matrices , their empirical eigenvalue distributions are supported on the unit circle @xmath19 and the empirical eigenvalue distribution of @xmath20 converges to the free multiplicative convolution @xmath13 , a probability measure supported on @xmath21 .
( we refer the reader to @xcite for an introduction to free probability theory .
we describe later the mechanics of calculating the free convolutions @xmath22 and @xmath16 . )
the fact that the empirical eigenvalue distribution of @xmath8 converges weakly to @xmath9 does not mean that all the eigenvalues of @xmath8 are close to the support of this measure .
there can be outliers , though they must not affect the limiting empirical eigenvalue distribution .
sometimes one can argue that these outliers must in fact exist .
for instance , the case when the rank @xmath23 of @xmath17 and its nonzero eigenvalues are fixed is investigated by benaych - georges and nadakuditi in @xcite . denote by @xmath24 these fixed eigenvalues .
of course , in this case @xmath25 is a point mass at @xmath26 , so the _ limiting _ behavior of the empirical eigenvalue distribution of @xmath8 is not affected by such a matrix @xmath0 .
more precisely , the empirical eigenvalue distribution of @xmath3 converges almost surely to the limiting spectral distribution @xmath7 of @xmath1 .
one can however detect , among the outlying eigenvalues of @xmath8 , the influence of the eigenvalues of @xmath0 above a certain critical threshold .
we use the notation @xmath27 for the eigenvalues of an @xmath28 matrix @xmath29 , repeated according to multiplicity .
the cauchy - stieltjes transform of a finite positive borel measure @xmath7 on @xmath10 is given by @xmath30 for @xmath31 outside the support of @xmath7 , and @xmath32 denotes the inverse of this function relative to composition .
when the support of @xmath7 is contained in the compact interval @xmath33 $ ] , the branch of the inverse function @xmath32 that satisfies @xmath34 is defined and real - valued on the interval @xmath35 , where @xmath36 the following result is proved in ( * ? ? ?
* theorems 2.1 and 2.2 ) .
[ thm : rrao ] 1 .
denote by @xmath37 and @xmath38 the infimum and supremum of the support of @xmath7 , respectively .
assume that the smallest and largest eigenvalues of @xmath1 converge almost surely to @xmath37 and @xmath38 , respectively .
then , almost surely for @xmath39 , @xmath40 similarly , almost surely for @xmath41 , @xmath42 2 .
fix @xmath43 such that @xmath44 . for
each @xmath2 define @xmath45 and let @xmath46 be a unit - norm eigenvector of @xmath47 associated to the eigenvalue @xmath48 .
then the following almost sure limits hold _ : _
@xmath49 and @xmath50 this result lies in the lineage of recent , and not so recent , works studying the influence of fixed rank additive or multiplicative perturbations on the extremal eigenvalues of classical random matrix models , the seminal paper being @xcite , where the so - called bbp phase transition was observed .
see also @xcite for sample covariance matrices , @xcite for deformed wigner models and @xcite for information - plus - noise models .
these investigations were first extended to perturbations of arbitrary rank in @xcite and @xcite for sample covariance matrices , in @xcite for deformed wigner models and in @xcite for information - plus - noise models .
it is pointed out in @xcite that the subordination function ( relative to the free additive convolution of a semicircular distribution with the limiting spectral distribution of the perturbation ) plays an important role in the fact that some eigenvalues of the deformed wigner model separate from the bulk .
note that in @xcite it is explained how the results of @xcite and @xcite in the setting of sample covariance matrices can also be described in terms of the subordination function related to the free multiplicative convolution of a marchenko - pastur distribution with the limiting spectral distribution of the multiplicative perturbation .
similar results have been obtained in ( * ? ? ?
* theorems 2.7 and 2.8 ) for multiplicative perturbations of the type @xmath12 , where @xmath51 is of fixed rank @xmath52 and @xmath53 . in this paper
we investigate the asymptotic behaviour of the eigenvalues and corresponding eigenvectors of the following models : * @xmath54 , where @xmath55 are deterministic , and @xmath56 is a haar - distributed unitary random matrix ; * @xmath57 where @xmath58 are deterministic , and @xmath56 is a haar - distributed unitary random matrix ; * @xmath59 , where @xmath60 are deterministic unitary matrices and @xmath56 is a haar - distributed unitary random matrix . in the first two models we assume that @xmath17 and @xmath18 have compactly supported limiting eigenvalue distributions @xmath6 and @xmath7 , respectively . in the third model we assume that the supports of @xmath6 and @xmath7 are not equal to the entire unit circle . a fixed number @xmath52 of eigenvalues of @xmath17the _
spikes_lay outside the support of @xmath6 for all @xmath61 , but all other eigenvalues of @xmath17 converge uniformly to the support of @xmath6 .
a similar hypothesis is made about @xmath18 and @xmath7 .
we answer the following questions : * when are some of the eigenvalues of @xmath62 almost surely located outside the support of the limiting spectral distribution @xmath9 ( respectively @xmath13 ) of @xmath62 ? in other words ,
when does the spectrum of @xmath62 have _ outliers _ almost surely ? * what is the behaviour of the eigenvectors corresponding to the outliers of @xmath62 ? when there are no spikes , it was shown in @xcite that , given a neighborhood @xmath63 of the support of @xmath9 , the eigenvalues of @xmath8 are almost surely contained in @xmath63 for large @xmath64 .
this paper extends the results of @xcite to perturbations of arbitrary rank , and it also extends the free probabilistic interpretation of outlier phenomena in terms of subordination functions as described in @xcite for deformations of wigner models .
indeed , the occurence and role of biane s subordination functions @xcite in the analysis of the interaction spikes / outliers is quite natural . we clarify this in the additive case through the following heuristics , leaving the precise statements to section [ sec : notation - and - statements ] .
let @xmath65 and @xmath66 be two free selfadjoint random variables in a tracial w*-probability space .
biane showed ( * ? ? ?
* theorem 3.1 ) that there exists an analytic self - map @xmath67 of the upper half - plane @xmath68 ( depending on @xmath69 ) so that @xmath70}\left[(z-({\bf a+b}))^{-1}\right]=(\omega(z)-{\bf a})^{-1 } , \quad z\in\mathbb c^+.\ ] ] here @xmath71}$ ] denotes the conditional expectation onto the von neumann algebra generated by @xmath65 and @xmath72 . the function @xmath73 is referred to as the _ subordination function_. ( this formula is a particular case of biane s result .
formula appears in this form in ( * ? ? ?
* appendix ) . )
the subordination function continues analytically via schwarz reflection through the complement in @xmath74 of the spectrum of @xmath75 .
if @xmath76 in distribution as @xmath77 , while a single eigenvalue @xmath78 , common to all of the matrices @xmath17 , stays outside the spectrum of @xmath65 , this eigenvalue will disappear in the large @xmath2 limit , in the sense that it will not influence the spectrum of @xmath65 .
thus , the analytic function @xmath79 will not be prevented _ a priori _ from taking the value @xmath78 . however , if relation were true with @xmath65 and @xmath66 replaced by @xmath17 and @xmath80 , respectively , and with the _ same _ subordination function @xmath73 , then any number @xmath81 in the domain of analyticity of @xmath73 so that @xmath82 must generate a polar singularity for the right - hand side of .
therefore , each such @xmath81 must generate a similar singularity for the term on the left - hand side of the same equality , thus necessarily producing an eigenvalue of @xmath83 . while this scenario is not true , we prove that an approximate version does hold .
namely , we show that a compression of the matrix @xmath84^{-1}+a_n\ ] ] to a subspace @xmath85 is close to @xmath86 , almost surely as @xmath77 .
this insight is crucial in our arguments .
it follows from our results that a remarkable new phenomenon occurs : a single spike of @xmath0 can generate asymptotically a finite , possibly arbitrarily large , set of outliers for @xmath8 .
this arises from the fact that the restriction to the real line of some subordination functions may be many - to - one , that is , with the above notation , the set @xmath87 may have cardinality strictly greater than 1 , unlike the subordination function related to free convolution with a semicircular distribution that was used in @xcite .
the case of multiplicative perturbations is based on similar ideas , with the subordination function @xmath73 replaced by its multiplicative counterparts ( * ? ? ?
* theorems 3.5 and 3.6 ) .
in addition to outliers , we investigate the corresponding eigenspaces of @xmath62 .
it turns out that the angle between these eigenvectors and the eigenvectors associated to the original spikes is determined by biane s subordination function , this time via its derivative .
the paper is organized as follows . in section 2 , we describe in detail the matrix models to be analysed , and state the main results of the paper . in section 3
we introduce free convolutions and the analytic transforms involved in their study .
we give the functional equations characterising the subordination functions . in section 4
we collect and prove a number of of auxiliary results , and in section 5 we prove the main results .
we denote by @xmath88 the upper half - plane , by @xmath89 the lower half - plane , and by @xmath90 the unit disc in @xmath91 .
the topological boundary of the unit disc is denoted by @xmath92 . for any vector subspace @xmath93 of @xmath94
, @xmath95 denotes the orthogonal projection onto @xmath93 .
@xmath96 stands for the set of @xmath97 matrices with complex entries , @xmath98 for the subset of invertible matrices , and @xmath99 for the unitary group .
the operator norm of a matrix @xmath29 is @xmath100 , its spectrum is @xmath101 its kernel is @xmath102 , its trace is @xmath103 and its normalized trace is @xmath104 .
the eigenvalues of a hermitian matrix @xmath29 are denoted @xmath105 and the probability measure @xmath106 is the empirical eigenvalue distribution of @xmath29 .
when @xmath29 is unitary , its eigenvalues are ordered decreasingly according to the size of their principal arguments in @xmath107 . if @xmath108 is a normal matrix , we denote by @xmath109 its spectral measure .
thus , if @xmath110 is a borel set , then @xmath111 is the orthogonal projection onto the linear span of all eigenvectors of @xmath29 corresponding to eigenvalues in @xmath112 .
the support of a measure @xmath81 on @xmath91 is denoted @xmath113 .
given any set @xmath114 , we define its @xmath115-neighbourhood by @xmath116 as long as there is no risk of confusion , the same notation will be used when @xmath117 and @xmath118 are subsets of @xmath21 .
open intervals in @xmath74 and open arcs in @xmath21 are denoted @xmath119 . as already seen in section 1 ,
the cauchy ( or cauchy - stieltjes ) transform of a finite positive borel measure @xmath81 on @xmath120 is an analytic function defined by @xmath121 iwe only consider finite measures @xmath81 supported @xmath74 or @xmath21 .
we denote by @xmath122 the reciprocal of @xmath123 that is , @xmath124 .
the moment generating function for @xmath81 is @xmath125 the @xmath126-transform of @xmath81 is defined as @xmath127 the relevant analytic properties of the transforms above are presented in subsections [ sec : boxplus][sec : boxtimest ] .
the _ free additive convolution _ of the borel probability measures @xmath6 and @xmath7 on @xmath74 is denoted by @xmath9 and the _ free multiplicative convolution _ of the borel probability measures @xmath6 and @xmath7 either on @xmath14 or on @xmath21 is denoted by @xmath13 .
given @xmath128 , denote by @xmath129 and @xmath130 the _ subordination functions _ corresponding to the free convolution @xmath9 .
they are known to be analytic on the complement of @xmath131 .
as the name suggests , they satisfy an analytic subordination property in the sense of littlewood : @xmath132 a similar result holds for the multiplicative counterparts of the subordination functions .
we have : @xmath133 where @xmath129 and @xmath130 are analytic on @xmath134 .
free convolutions are defined in subsections [ sec : boxplus][sec : boxtimest ] , and the subordination functions are defined via functional equations in subsections [ subsubsec : eq+][subsubsec : eqt ] in the following three subsections we describe our models and state the main results . to avoid dealing with too many special cases , we make the following technical assumption , which will be in force for the remainder of the paper , except for remark [ atoms ] .
@xmath135 under this assumption , the subordination functions extend continuously to the boundary ( see lemmas [ lem : extension ] , [ lem : extensionx ] , and [ lem : extensiont ] ) .
our results however remain substantially valid without this assumption , and we discuss in remark [ atoms ] the relevant modifications . here are the ingredients for constructing the additive matrix model @xmath54 : * two compactly supported borel probability measures @xmath6 and @xmath7 on @xmath136 . * a positive integer @xmath137 and fixed real numbers @xmath138 which do not belong to @xmath139 . * a sequence @xmath140 of deterministic hermitian matrices of size @xmath28 such that * * @xmath141 converges weakly to @xmath6 as @xmath77 ; * * for @xmath142 and @xmath143 , the sequence @xmath144 satisfies @xmath145 * * the eigenvalues of @xmath0 which are not equal to some @xmath146 converge uniformly to @xmath139 as @xmath77 , that is @xmath147 * a positive integer @xmath148 and fixed real numbers @xmath149 which do not belong to @xmath150 . * a sequence @xmath151 of deterministic hermitian matrices of size @xmath28 such that * * @xmath152 converges weakly to @xmath7 as @xmath77 ; * * for @xmath153 and @xmath154 ,
the sequence @xmath155 satisfies @xmath156 * * the eigenvalues of @xmath1 which are not equal to some @xmath157 converge uniformly to @xmath150 as @xmath77 . * a sequence @xmath158 of unitary random matrices such that the distribution of @xmath4 is the normalized haar measure on the unitary group @xmath5 .
it is known from @xcite that the asymptotic eigenvalue distribution of @xmath62 is @xmath9 . in the following statement
we take advantage of the fact , discussed later , that the functions @xmath129 and @xmath130 extend continuously to the real line .
the points in @xmath159 satisfying @xmath160 are isolated but they may accumulate to @xmath161 .
we denote by @xmath162 and @xmath163 the projection onto the space generated by the eigenvectors corresponding to the spikes of @xmath17 and @xmath18 , respectively .
these can also be written as @xmath164 in terms of the spectral measures of @xmath17 and @xmath18 .
[ main+ ] with the above notation , set @xmath165 , @xmath166\cup\left[\bigcup_{j=1}^q \omega_2^{-1}(\{\tau_j\})\right],\ ] ] and let @xmath167 be the subordination functions satisfying ( [ subord1 ] ) .
the following results hold almost surely for large @xmath2 : 1 . given @xmath168 , we have @xmath169 2 .
fix a number @xmath170 , let @xmath168 be such that @xmath171 , and set @xmath172 , @xmath173 . then @xmath174 3 .
with @xmath81 and @xmath115 as in part @xmath175 , we have @xmath176 and @xmath177 4 . with @xmath81 and @xmath115 as in part @xmath175 ,
suppose in addition that @xmath178 .
then @xmath179 analogously , if @xmath180 , then @xmath181 in case @xmath182 and @xmath183 , the result of ( 3 ) above implies the following .
let @xmath184 be an orthonormal basis of @xmath185 . then , almost surely for large @xmath2 , we have @xmath186 and @xmath187 for @xmath188 , where @xmath189 is the usual kronecker symbol .
thus , assertion ( 4 ) is a strenghtening of ( 3 ) in the special case @xmath190 .
here are the ingredients for constructing the multiplicative model @xmath191 : * two compactly supported borel probability measures @xmath6 and @xmath7 on @xmath192 . * a positive integer @xmath137 , and fixed positive numbers @xmath193 which do not belong to @xmath139 . * a sequence @xmath140 of deterministic nonnegative matrices of size @xmath28
such that * * @xmath141 converges weakly to @xmath6 as @xmath77 , * * for @xmath142 and @xmath194 the sequence @xmath195 satisfies @xmath196 * * the eigenvalues of @xmath0 which are not equal to some @xmath146 converge uniformly to @xmath139 as @xmath77 .
* a positive integer @xmath148 , and fixed positive numbers @xmath197 which do not belong to @xmath150 . * a sequence @xmath151 of deterministic nonnegative matrices of size @xmath28 such that * * @xmath152 converges weakly to @xmath7 as @xmath77 , * * for @xmath153 and @xmath198 the sequence @xmath199 satisfies @xmath200 * * the eigenvalues of @xmath1 which are not equal to some @xmath157 converge uniformly to @xmath150 as @xmath77 . * a sequence @xmath158 of random matrices such that the distribution of @xmath4 is the normalized haar measure on the unitary group @xmath5 .
it is known from @xcite that the asymptotic eigenvalue distribution of @xmath62 is @xmath13 .
the projections @xmath162 and @xmath163 used in the following statement were defined in .
[ mainx+ ] with the above notations , let @xmath167 be the subordination functions satisfying _ ( [ subord1 ] ) _ , set @xmath201 , @xmath202 , @xmath203 , and @xmath204\cup\left[\bigcup_{j=1}^q v_2^{-1}(\{1/\tau_j\})\right].\ ] ] the following results hold almost surely for large @xmath2 : 1
. given @xmath168 we have @xmath169 2 . fix a positive number @xmath170 , let @xmath168 be such that @xmath205 and set @xmath206 , @xmath207
. then @xmath174 3 . with @xmath81 and @xmath115 as in part @xmath175 , we have @xmath208 and @xmath209 4 . with @xmath81 and @xmath115 as in part @xmath175 ,
suppose in addition that @xmath178 .
then @xmath210 analogously , if @xmath180 , then @xmath211 finally , we describe the ingredients for the construction of the multiplicative matrix model @xmath59 with unitary @xmath17 and @xmath18 : * two borel probability measures @xmath6 and @xmath7 on @xmath21 with nonzero first moments such that @xmath212 . * a positive integer @xmath137 and fixed complex numbers @xmath213 which do not belong to @xmath139 and such that @xmath214 * a sequence @xmath140 of deterministic unitary matrices of size @xmath28 such that * * @xmath141 converges weakly to @xmath6 as @xmath77 , * * for @xmath142 and @xmath194 the sequence @xmath195 satisfies @xmath215 * * the eigenvalues of @xmath0 which are not equal to some @xmath146 converge uniformly to @xmath139 as @xmath77 . * a positive integer @xmath148 and fixed complex numbers @xmath216 which do not belong to @xmath150 and such that @xmath217 * a sequence @xmath151 of deterministic unitary matrices of size @xmath28 such that * * @xmath152 converges weakly to @xmath7 as @xmath77 , * * for @xmath153 and @xmath198 the sequence @xmath199 satisfies @xmath218 * * the eigenvalues of @xmath1 which are not equal to some @xmath157 converge uniformly to @xmath150 as @xmath77 . * a sequence @xmath158 of random matrices such that the distribution of @xmath4 is the normalized haar measure on the unitary group @xmath5 .
it is known from @xcite that the asymptotic eigenvalue distribution of @xmath62 is @xmath13 .
when @xmath219 and @xmath168 , the interval @xmath220 consists of those numbers in @xmath21 whose argument differs from @xmath221 by less than @xmath115 . with this convention , theorem [ mainx+ ] holds _ verbatim _ in the unitary case as well .
free convolutions arise as natural analogues of classical convolutions in the context of free probability theory . for two borel probability measures @xmath6 and @xmath7 on the real line , one defines the free additive convolution @xmath9 as the distribution of @xmath222 , where @xmath37 and @xmath38 are free self - adjoint random variables with distributions @xmath6 and @xmath7 , respectively . similarly ,
if both @xmath128 are supported on @xmath14 or on @xmath223 , their free multiplicative convolution @xmath13 is the distribution of the product @xmath224 , where , as before , @xmath37 and @xmath38 are free , positive in the first case , unitary in the second , random variables with distributions @xmath6 and @xmath7 , respectively .
the product @xmath224 of two free positive random variables is usually not positive , but it has the same moments as the positive random variables @xmath225 and @xmath226 we refer to @xcite for an introduction to free probability theory and to @xcite for the definitions and main properties of free convolutions . in this section , we recall the analytic approach developed in @xcite to calculate the free convolutions of measures , as well as the analytic subordination property @xcite and related results .
recall from the definition of the cauchy - stieltjes of a finite positive borel measure @xmath6 on the real line : @xmath227 this function maps @xmath228 to @xmath229 and @xmath230 .
conversely , any analytic function @xmath231 for which @xmath232 is finite is of the form @xmath233 for some finite positive borel measure @xmath6 on @xmath74 .
when @xmath6 has compact support , the function @xmath234 is also analytic at @xmath235 and @xmath236 ( see ( * ? ?
* chapter 3 ) for these results ) .
the measure @xmath6 can be recovered from its cauchy - stieltjes transform as a weak limit @xmath237 ( this holds for signed measures as well . )
the density of ( the absolutely continuous part of ) @xmath6 relative to lebesgue measure is calculated as @xmath238 for almost every @xmath239 relative to the lebesgue measure .
in particular , @xmath240 can be described as the set of those points @xmath241 with the property that @xmath242 can be continued analytically to an open interval @xmath243 such that @xmath244 is real - valued .
on the other hand , @xmath245 almost everywhere relative to the singular part of @xmath6 .
indeed , these facts follow from the straightforward observation that @xmath246 @xmath247 , is the poisson integral of @xmath6 .
see @xcite for these aspects of harmonic analysis .
it is often convenient to work with the reciprocal cauchy - stieltjes transform @xmath248 , which defines an analytic self - map of the upper half - plane .
this function enjoys the following properties : * for any @xmath249 , @xmath250 .
if equality holds at one point of @xmath251 , then it holds at all points , and @xmath252 . * in particular
, the function @xmath253 is a self - map of @xmath251 unless @xmath6 is a point mass , in which case @xmath254 is a real constant .
* if @xmath6 is compactly supported , there exist a real number @xmath255 and a finite positive borel measure @xmath81 on @xmath74 with @xmath113 included in the convex hull of @xmath139 such that @xmath256 conversely , if @xmath257 extends to an analytic real - valued function to the complement in @xmath74 of a compact set , and if @xmath258 , then there exists a compactly supported positive borel measure @xmath6 on @xmath74 so that @xmath259
. the value @xmath260 is determined by @xmath261 . *
if @xmath262 and @xmath81 is as in , we have @xmath263 and @xmath264 .
equation is a special case of the nevanlinna representation of analytic self - maps of the upper half - plane ( * ? ? ?
* chapter 3 ) : @xmath265 where @xmath266 and @xmath267 is a positive finite borel measure on @xmath74 .
we identify @xmath268 , @xmath269 , @xmath270 if @xmath271 , then reduces to , with @xmath272 and @xmath273 .
the cauchy - stieltjes transform of a compactly supported probability measure @xmath6 is conformal in the neighborhood of @xmath235 , and its functional inverse @xmath274 is meromorphic at zero with principal part @xmath275 .
the @xmath276-transform @xcite of @xmath6 is the convergent power series defined by @xmath277 the free additive convolution of two compactly supported probability measures @xmath6 and @xmath7 is another compactly supported probability measure characterized by the identity @xmath278 satisfied by these convergent power series .
recall from the definition of the moment - generating function of a borel probability measure @xmath6 on @xmath14 : @xmath279 this function is related to the cauchy - stieltjes transform of @xmath6 via the relation @xmath280 it satisfies the following properties , for which we refer to ( * ? ? ?
* section 6 ) : * @xmath281 . *
@xmath282 and @xmath283 in addition , @xmath284 * in particular ,
if @xmath139 is compact and not equal to @xmath285 , then @xmath286 is injective on some neighbourhood of zero in @xmath120 .
* @xmath286 is injective on @xmath287 .
it is often convenient to work with the eta transform , or boolean cumulant function , @xmath288 it inherits from @xmath289 the following properties : * @xmath290 for all @xmath249 , where @xmath291 takes values in @xmath292 on @xmath251 .
moreover , if equality holds for one point in @xmath251 , it holds for all points in @xmath251 , and @xmath293 for any @xmath249 .
* @xmath294 and @xmath295 in particular , if @xmath139 is compact and different from @xmath285 , then @xmath296 is injective on some neighbourhood of zero in @xmath120 . *
if @xmath297 , @xmath296 is strictly increasing from @xmath298 $ ] to @xmath299 $ ] , where @xmath300 should be replaced by @xmath301 if @xmath302 .
moreover , @xmath296 is injective on @xmath287 . *
conversely , if an analytic function @xmath303 satisfies @xmath304 for all @xmath249 and @xmath305 , then @xmath306 for some borel probability measure on @xmath14 ( * ? ? ?
* proposition 2.2 ) .
the @xmath307-transform @xcite of a compactly supported measure @xmath297 is the convergent power series defined by @xmath308 where @xmath309 is the inverse of @xmath296 relative to composition .
the free multiplicative convolution of two compactly supported probability measures @xmath310 is another compactly supported probability measure characterized by the identity @xmath311 in a neighbourhood of @xmath26 .
the analytic transforms involved in the study of multiplicative free convolution on @xmath21 are formally the same ones as in subsection [ sec : boxtimes+ ] , but their analytical properties are different .
thus , @xmath312 it satisfies @xmath313 for all @xmath314 .
we work almost exclusively with the eta transform , or boolean cumulant function , @xmath288 the following properties of @xmath296 are relevant to our study : * for any @xmath315 , we have @xmath316 .
if equality holds at one point in @xmath317 , it holds at all points in @xmath318 , and @xmath319 for any @xmath320 .
* @xmath321 and @xmath322 in particular @xmath296 is injective on a neighbourhood of zero in @xmath120 if and only if @xmath323 .
* the function @xmath296 continues via schwarz reflection through the set @xmath324 , that is @xmath325 * for almost all points @xmath326 with respect to the absolutely continuous part of @xmath6 ( relative to the haar measure on @xmath21 ) , the nontangential limit of @xmath296 at @xmath239 belongs to @xmath318 , and for almost all points @xmath326 in the complement of the support of the absolutely continuous part of @xmath6 , the nontangential limit of @xmath296 at @xmath239 belongs to @xmath21 .
moreover , if @xmath6 has a singular component , then for almost all points @xmath326 with respect to this component , the nontangential limit of @xmath296 at @xmath239 equals one . *
conversely , if an analytic function @xmath327 satisfies @xmath328 , then @xmath306 for a unique borel probability measure on @xmath21 ( * ? ? ?
* proposition 3.2 ) .
when @xmath323 , we define the @xmath307-transform @xcite of @xmath6 as the convergent power series @xmath329 again , the free multiplicative convolution of two probability measures @xmath6 and @xmath7 with nonzero first moments is another probability measure with nonzero first moment characterized by the identity @xmath311 in a neighbourhood of @xmath26 .
if both of @xmath6 and @xmath7 have zero first moment , then @xmath13 is the haar ( uniform ) distribution on @xmath21 , see @xcite . from now on
, we always assume that all our probability measures on @xmath21 have nonzero first moments .
the analytic subordination phenomenon for free convolutions , as described in and , was first noted by voiculescu in @xcite for free additive convolution of compactly supported probability measures .
later , biane @xcite extended the result to free additive convolutions of arbitrary probability measures on @xmath74 , and also found a subordination result for multiplicative free convolution .
more importantly , he proved the stronger result ( see heuristics in the introduction ) that the conditional expectation of the resolvent of a sum or product of free random variables onto the algebra generated by one of them is in fact also a resolvent . in @xcite , voiculescu deduced this property from the fact that such a conditional expectation is a coalgebra morphism for certain coalgebras , and through this observation he extended the subordination property to free convolutions of operator - valued distributions . for our purposes ,
considerably less than that is required : we essentially need only complex analytic properties of these functions .
given borel probability measures @xmath6 and @xmath7 on @xmath74 , there exist two unique analytic functions @xmath330 such that 1 .
@xmath331 , @xmath203 ; 2 .
@xmath332 3 .
in particular ( see @xcite ) , for any @xmath333 so that @xmath129 is analytic at @xmath31 , @xmath334 is the attracting fixed point of the self - map of @xmath251 defined by @xmath335 a similar statement , with @xmath128 interchanged , holds for @xmath130 .
we note that implies that the functions @xmath167 continue analytically accross an interval @xmath336 such that @xmath337 and @xmath338 are real valued if and only if the same is true for @xmath339 . for the sake of providing an intrinsic characterization of the correspondence betweem spikes and outliers , we formalize and slightly strenghten this remark in the following lemma . here
we use the functions @xmath340 defined by ( [ h ] ) .
[ lem : extension ] consider two compactly supported borel probability measures @xmath6 and @xmath7 , neither of them a point mass .
then the subordination functions @xmath129 and @xmath130 have extensions to @xmath341 with the following properties : 1 .
@xmath342 are continuous .
if @xmath343 then the functions @xmath129 and @xmath130 continue meromorphically to a neighborhood of @xmath239 , @xmath344 , and @xmath345 . if @xmath346 , then @xmath347 , and if @xmath348 , then @xmath349 .
conversely , suppose that @xmath129 continues meromorphically to a neighbourhood of a point @xmath350 and that @xmath351 when @xmath352 for some @xmath353 .
if @xmath354 , then @xmath239 is an isolated atom for @xmath9 . in the context of part ( b ) of the above lemma
, we note that @xmath254 is analytic around infinity , and @xmath355 .
part ( a ) was proved in ( * ? ? ?
* theorem 3.3 ) .
fix @xmath356 .
equation indicates that @xmath129 and @xmath130 must take real values on @xmath357 for some @xmath358 .
schwarz reflection implies that @xmath129 and @xmath130 have meromorphic continuations with real values on @xmath74 accross the corresponding intervals .
the relation @xmath359 shows that the limit @xmath360 is real for @xmath361 and therefore @xmath362 by the stieltjes inversion formula .
in particular , @xmath363 . to conlcude the proof of @xmath364 ,
suppose that @xmath346 .
it follows from in conjunction with items ( c ) and ( d ) of subsection [ sec : boxplus ] that @xmath365 so that @xmath130 is analytic at @xmath239 .
the statement for @xmath348 follows by symmetry .
finally , suppose that the hypotheses of ( c ) is satisfied .
it was observed in @xcite that @xmath366 is also real for @xmath361 .
( indeed , if @xmath367 , relation implies @xmath368 and therefore @xmath369 .
this relation can only hold when @xmath7 is a point mass , a case which we excluded . ) now , implies that @xmath339 is continuous and real - valued on @xmath357 , and this yields the desired conclusion via the stieltjes inversion formula . given borel probability measures @xmath370 on @xmath14 , there exist two unique analytic functions @xmath371 so that 1
. @xmath372 for @xmath249 and @xmath203 ; 2 .
@xmath373 3 .
in particular ( see @xcite ) , for any @xmath333 so that @xmath129 is analytic at @xmath31 , the point @xmath374 is the attracting fixed point of the self - map of @xmath375 defined by @xmath376 a similar statement , with @xmath128 interchanged , holds for @xmath130 . a version of lemma [ lem : extension ]
holds for multiplicative free convolution on @xmath14 .
since the proof is similar to the proof of lemma [ lem : extension ] and of lemma [ lem : extensiont ] below , we omit it .
item ( a ) appears in the proof of ( * ? ? ? * theorem 3.2 ) .
[ lem : extensionx ] consider two compactly supported borel probability measures @xmath128 on @xmath14 , neither of them a point mass .
then the restrictions of the subordination functions @xmath129 and @xmath130 to @xmath251 have extensions to @xmath341 with the following properties : 1 .
@xmath377 are continuous .
2 . if @xmath378 then the functions @xmath129 and @xmath130 continue analytically to a neighborhood of @xmath239 , @xmath379 , and @xmath380 . given borel probability measures @xmath128 on @xmath21 with nonzero first moments , there exist unique analytic functions @xmath381 so that 1 . @xmath382 , @xmath203 , @xmath315 ; 2 .
@xmath383 3 .
in particular ( see @xcite ) , if @xmath384 and @xmath129 is analytic at @xmath31 , then the point @xmath374 is the attracting fixed point of the map @xmath385 a similar statement , with @xmath128 interchanged , holds for @xmath130 .
[ lem : extensiont ] consider two borel probability measures @xmath128 on @xmath21 with nonzero first moments , neither of them a point mass .
suppose that @xmath386 .
then the subordination functions @xmath129 and @xmath130 have extensions to @xmath387 with the following properties : 1 .
@xmath388 are continuous .
2 . if @xmath389 then the functions @xmath129 and @xmath130 continue analytically to a neighborhood of @xmath239 , @xmath390 , and @xmath391 .
part ( a ) can be found in ( * ? ? ?
* theorem 3.6 ) .
fix @xmath392 . equation coupled with items ( d ) and ( e ) of subsection [ sec : boxtimest ] indicate clearly that @xmath167 must take values in @xmath21 at least a.e . on a neighbourhood of @xmath239 . as proved in ( * ? ? ?
* proposition 1.9 ( a ) ) if , say , @xmath129 does not reflect analytically through a neighbourhood of @xmath239 , then for any @xmath393 the set of nontangential limits @xmath394 of @xmath129 around @xmath239 is dense in @xmath21 .
as @xmath395 is nonempty , many of these limits will fall in the domain of analyticity of @xmath296 .
in particular , we may choose an arbitrary interval @xmath396\}$ ] strictly included in the domain of analyticity of @xmath296 and we will be able to find points @xmath397 tending to @xmath326 so that @xmath398 obviously , in that case any limit point of @xmath399 will still belong to @xmath400 , and hence be in the domain of analyticity of @xmath296 .
pick such a limit point @xmath401 .
note that , as a trivial consequence of the julia - carathodory theorem ( * ? ? ?
* chapter i , exercises 6 and 7 ) , @xmath402 for any @xmath403 in the domain of analyticity of @xmath296 , and thus @xmath404 , which implies that @xmath296 is conformal on a neighbourhood @xmath405 of @xmath401 ( in @xmath406 . now recall that @xmath407 letting @xmath408 go to infinity ( along a subsequence , if necessary ) , and recalling that @xmath409 we obtain @xmath410 .
both functions are analytic around the two respective points from @xmath21 , so the conformality of @xmath296 on @xmath405 allows us to find @xmath411 as a neighbourhood of @xmath412 on which the compositional inverse @xmath413 can be defined .
we write @xmath414 on some convex neighbourhood @xmath415 of @xmath239 which is small enough so that @xmath416 ( the existence of @xmath415 is guaranteed by the continuity of @xmath417 around @xmath239 ) .
pick points @xmath418 so that @xmath419
. clearly @xmath420 , so that for all @xmath421 large enough , @xmath422 . pick a piecewise linear path going consecutively through the @xmath423 s , so that , by the convexity of @xmath415 , this path stays in @xmath415 and necessarily converges to @xmath239 .
for any @xmath31 in this path , we have @xmath424 which , by analytic continuation and analyticity of @xmath414 on @xmath415 , forces @xmath129 to be analytic on @xmath415 , providing a contradiction .
thus , @xmath129 extends analytically through @xmath239 .
the same argument shows that @xmath130 extends analytically around @xmath239 .
the last statement of ( b ) above follows again from the simple remark that @xmath425 unlike the case of free additive convolution , the functions @xmath129 and @xmath130 are bounded on @xmath318 and hence do not have pole singularities on @xmath21 .
the proofs of our main results will be based largely on both scalar- and matrix - valued analytic function methods , as well as on some elementary results from operator theory .
we start by collecting some results which apply to both additive and multiplicative models .
we use the notation introduced in section [ sec : notation - and - statements ] . the following convergence result for sequences of nevanlinna - type functions
is necessary in the analysis of eigenvectors corresponding to outliers .
@xmath426 denotes the space of complex - valued continuous functions on a topological space @xmath427 .
we use the notation @xmath428 for the total variation measure of a signed measure @xmath81 on @xmath74 .
that is , @xmath429 , where @xmath430 is the hahn decomposition of @xmath81 .
the total variation of @xmath81 is @xmath431 .
[ conv ] let @xmath432 be a family of signed measures on @xmath74 satisfying the following properties : * there exists @xmath433 such that @xmath434 $ ] for all @xmath61 ; * @xmath435 ; * @xmath436 in the weak@xmath437-topology as @xmath77 , that is , @xmath438).\ ] ] then there exists a sequence @xmath439 converging to zero , independent of @xmath31 , such that @xmath440 fix @xmath168 .
it is known from @xcite that @xmath441 for any fixed @xmath247 .
we prove that there exists @xmath442 such that @xmath443 the proof is then completed by setting @xmath444 indeed , as noted in , there exists @xmath445 such that @xmath446 on the other hand , for @xmath447 $ ] we have @xmath448)\frac{(\im z)^2}{|z|^2+\sup_{n\in\mathbb n}{\|\rho_n\|}}\\ & \leq & \im z\frac{\|\rho_n\|}{|z|^2+\sup_{n\in\mathbb n}{\|\rho_n\| } } < \im z\leq\varepsilon.\end{aligned}\ ] ] we conclude that @xmath449 , as desired .
an analogous result holds for @xmath21 .
[ convt ] let @xmath450 be a family of analytic self - maps of the unit disc such that @xmath451 and the limit @xmath452 exists for all @xmath453 .
then there exists a sequence @xmath439 converging to zero , independent of @xmath31 , such that @xmath454 there exists a unique probability measure @xmath455 on @xmath74 such that @xmath456 this is seen by applying to the map @xmath457 similarly , @xmath458 since @xmath459 is a bijection from @xmath74 to @xmath460 , we have @xmath461 thus , @xmath462 provided that @xmath463 .
on the other hand , montel s theorem implies that the convergence @xmath464 is uniform on compact subsets of @xmath318 , and thus there exists @xmath442 such that @xmath465 the sequence @xmath466 satisfies the conclusions of the lemma . the following lemma from ( * ? ? ? * appendix ) is proved using ideas from @xcite .
we use the notation @xmath467 for the space of infinitely differentiable , compactly supported functions @xmath468 .
[ ht ] let @xmath469 be an analytic function on @xmath470 which satisfies @xmath471 for some numbers @xmath472 , @xmath473 , and polynomial @xmath474 with nonnegative coefficients .
for every @xmath475 there exists a constant @xmath476 independent of @xmath469 such that @xmath477 we also record a result from ( * ? ? ?
* lemma 6.3 ) on the boundary behaviour of a certain poisson kernel convolution ( recall and the comments following it ) .
we use @xmath478 to denote the expectation .
if @xmath479 is a continuous function and @xmath480 is a selfadjoint matrix , then @xmath481 is constructed using the continuous functional calculus . [ approxcauchy ] given a deterministic @xmath28 hermitian matrix @xmath482 , a random @xmath28 hermitian matrix @xmath62 , and a continuous function @xmath483 with compact support , we have @xmath484 \right]= \lim_{y\downarrow 0 } \frac{1}{\pi } \im \int_\mathbb r \mathbb{e } \left [ { \rm tr}_n [ \left(x_n-(t+iy)i_n\right)^{-1}c_n ] \right]h(t)\ , dt.\ ] ] an essential ingredient in our analysis is the resolvent of @xmath17 and of the matrices @xmath62 ( depending on the model considered , @xmath54 , @xmath485 or @xmath59 ) .
we denote by @xmath486 the resolvent of @xmath62 .
it is a random matrix - valued rational function with poles in @xmath74 for the first two models and @xmath21 for the third .
for the first two models , it has the following properties : 1 .
in particular , @xmath488 is selfadjoint if @xmath489 2 .
@xmath490 is analytic at @xmath235 , and @xmath491 , where @xmath492 denotes the @xmath28 identity matrix .
the limit is in the norm topology of @xmath493 , where @xmath494 denotes the haar measure on @xmath5 .
3 . with the notation @xmath495 and @xmath496 for the real and imaginary parts of @xmath497 ,
respectively , we have @xmath498 this last quantity is uniformly bounded below in @xmath2 for @xmath31 in any fixed compact set @xmath499 .
in particular , if @xmath500 is such that @xmath501 , @xmath502\ge\frac{\im z}{|z|^2 + 2|\re z|c+c^2}i_n.\ ] ] for the unitary model , a slightly different property is needed . 1 . if @xmath315 , we have @xmath503 , and thus @xmath504 . therefore
@xmath505>\frac12i_n,\quad z\in\mathbb d.\ ] ] ( this observation uses the fact that @xmath62 is unitary , and hence normal . ) the following lemma is a fairly straightforward generalization of a result of hurwitz .
a similar result appears in @xcite . in the statement
, we use @xmath506 to denote the subset of @xmath507 consisting of all points at distance strictly less than @xmath508 from @xmath117 . in the special case @xmath509
, we write @xmath510 instead of @xmath506 .
[ alt - benaych - rao ] let @xmath507 be a simple analytic curve in @xmath511 , let @xmath512 be compact , and let @xmath23 be a positive integer . consider an analytic function @xmath513 such that @xmath514 is diagonal for each @xmath515 , @xmath516 and @xmath517 has only simple zeros , all of which are contained in @xmath518 , @xmath519 .
fix @xmath353 such that @xmath520 has no zeroes on the boundary of @xmath506 relative to @xmath507 , and let @xmath521 be a list of those points @xmath522 for which @xmath514 is not invertible .
suppose that there exist positive numbers @xmath523 and analytic maps @xmath524 , @xmath525 such that : 1 .
@xmath526 2 .
@xmath527 is invertible for @xmath528 and @xmath529 and 3 .
@xmath530 converges to @xmath531 uniformly on compact subsets of @xmath532 .
then : 1 . @xmath533
equals the order of @xmath534 as a zero of @xmath535 2 .
given @xmath168 such that @xmath536 there exists an integer @xmath537 such that for @xmath538 , we have 1 . counting multiplicities , @xmath539 has exactly @xmath540 zeroes in @xmath541 , @xmath542 , and 2 .
@xmath543 assertion ( i ) is obvious .
the functions @xmath544 converge to @xmath545 uniformly on compact subsets of @xmath546 . the theorem of hurwitz ( see ( * ? ? ?
* kapitel 8.5 ) ) guarantees that , for sufficiently large @xmath2 , @xmath547 has ( counting multiplicities ) exactly as many zeros as @xmath548 in @xmath549 .
all the zeros of @xmath547 were assumed to be in @xmath507 and therefore these zeros cluster around @xmath550 in the following sense : for any given @xmath168 , there exists an @xmath551 so that @xmath552 when @xmath553
. when @xmath168 is small enough , there are ( counting multiplicities ) exactly @xmath554 zeros of @xmath547 in @xmath555 .
later , we apply this lemma to @xmath556 and @xmath557 , in order to control the behaviour of functions related to the resolvent @xmath490 . + next
, we collect some facts about matrix functions and maps on matrix spaces which commute with the operation of conjugation by unitary matrices .
first , an analog of the nevanlinna representation for matrix - valued functions ( * ? ? ?
* section 5 ) .
let @xmath558 be fixed and let @xmath559\to m_n(\mathbb c)$ ] be an analytic function .
assume that @xmath560 is nonnegative definite for @xmath561 , and @xmath562 for @xmath563 $ ] .
then @xmath531 can be represented as @xmath564}\frac{d\rho(t)}{z - t},\quad z\in\mathbb{c}\setminus[-m , m],\ ] ] where @xmath480 is a selfadjoint matrix , @xmath565 , and @xmath81 is a measure with values in @xmath566 such that @xmath567 for every borel set @xmath568 . observe that @xmath569 the norm of such a function can obviously be estimated as @xmath570 the specific situation we have in mind is as follows .
let @xmath29 be a random selfadjoint matrix in @xmath566 such that @xmath571 almost surely .
pick @xmath572 such that @xmath573 ( that is , @xmath574 is positive definite ) .
the matrix @xmath575 $ ] is analytic in @xmath31 , it is invertible for @xmath576,\ ] ] and it is selfadjoint for @xmath577.\ ] ] moreover , @xmath578 < 0,\quad z\in\mathbb{c}^{+}.\ ] ] it follows that the function @xmath579)^{-1}$ ] satisfies the properties required for it to have a representation of the form .
the matrices @xmath580 and @xmath581 are easily determined .
indeed , we have @xmath582)^{-1}&=&\im b-\varepsilon\mathbb{e}[x-\re b]-\varepsilon^{2}\left[\mathbb{e}[x(\im b)^{-1}x]\right.\\ & & \mbox{}-\left.\mathbb{e}[x](\im b)^{-1}\mathbb e[x]\right]+o(\varepsilon^{3})\end{aligned}\ ] ] as @xmath583 . substituting @xmath584
, we obtain @xmath585+\re b - \frac{\mathbb{e}[(x-\mathbb{e}[x])(\im b)^{-1}(x-\mathbb{e}[x])]}{z}+o\left(\frac{1}{z^{2}}\right)\ ] ] as @xmath586 .
this yields @xmath587+\re b,\quad \textbf{$b=\im b$},\quad \text{and}\quad \rho(\mathbb{r})=\mathbb{e}[(x-\mathbb{e}[x])(\im b)^{-1}(x-\mathbb{e}[x])].\ ] ] we are mostly , but not exclusively , interested in the case @xmath588 .
these observations apply to the variables @xmath62 from our models .
we begin with @xmath589 , where @xmath590 and @xmath591 are any sequences of deterministic real diagonal matrices of size @xmath28 with uniformly bounded norms and limiting distributions @xmath6 and @xmath592 , respectively . as before , @xmath593 . more generally , if @xmath594 satisfies @xmath595 , then @xmath596 .
[ l ] the function @xmath597 $ ] takes values in @xmath598 whenever @xmath595 .
moreover , @xmath599^{-1}\ge\im b\text { and } \|\mathbb e[r_{n}(b)]^{-1}\|\le\|b\|+c_1+{4c_2}{\|(\im b)^{-1}\|},\quad\im b>0,\ ] ] where @xmath600 , and @xmath601^{2})$ ]
. in particular , @xmath602^{-1}\ge\im zi_n\text { and } \vert\mathbb{e}[r_{n}(z)]^{-1}\vert\le|z|+c_{1}+\frac{4c_{2}}{|\im z|},\quad z\in\mathbb{c}^+.\label{borne+}\ ] ] the relation @xmath603^{-1}\ge\im b$ ] follows from ( * ? ? ?
* remark 2.5 ) .
the second inequality follows immediately from the observations preceding the lemma , and from the fact that for any deterministic matrix @xmath604 , @xmath605-\mathbb{e}[u_{n}^{*}b_{n}u_{n}]z\mathbb{e}[u_{n}^{*}b_{n}u_{n}]}\\ & = ( { \rm tr}_{n}(b_{n}^{2})-[{\rm tr}_{n}(b_{n})]^{2})\left(\frac{n^2}{n^2 - 1}\text{tr}_n(z)i_n- \frac{1}{n^2 - 1}z\right).\qedhere\end{aligned}\ ] ] in some situations it is convenient to see how @xmath606 $ ] depends on @xmath17 ; recall that @xmath54 .
this is achieved to some extent by the following lemma ( see also @xcite ) .
[ bicommutant ] fix a matrix @xmath607 .
let @xmath572 be such that @xmath608 is invertible for every @xmath609 , consider the random matrix @xmath610 and its expected value @xmath611 $ ] . then : 1 . for every @xmath612 we have @xmath613.\label{eq : first - commutation}\ ] ]
if @xmath614 is invertible , we also have @xmath615g(b)^{-1}.\nonumber\end{aligned}\ ] ] 2 .
@xmath616 , where @xmath617 denotes the double commutant of @xmath38 in @xmath566 .
the conclusion of item ( 2 ) of the above lemma applies to any complex differentiable map @xmath548 defined on an open set in @xmath618 with the property that @xmath619 for all @xmath620 .
the analytic function @xmath621 $ ] is defined in an open set containing the selfadjoint matrices .
moreover , left invariance of the haar measure implies that @xmath622 is constant on the selfadjoint matrices .
since the selfadjoint matrices form a uniqueness set for analytic functions , we deduce that @xmath622 is constant in a neighborhood of the selfadjoint matrices .
in particular , given @xmath612 , the function @xmath623\ ] ] does not depend on @xmath115 for small @xmath624 .
differentiation at @xmath625 yields the identity @xmath626=0.\ ] ] using now the fact that @xmath627 and @xmath628 , we obtain @xmath629=0\ ] ] which is equivalent to because @xmath630=g(b)y$ ] and @xmath631=yg(b)$ ] . to prove the second identity in ( 1 ) ,
observe that @xmath632=\mathbb{e}[(r(b)-g(b))(yb - by)(r(b)-g(b))]+g(b)(yb - by)g(b)\ ] ] so that implies @xmath633.\ ] ] relation is now obtained multiplying this relation by @xmath634 on both sides .
to verify ( 2 ) , we need to show that @xmath614 commutes with any matrix @xmath635 .
this follows immediately from ( [ eq : first - commutation ] ) .
the preceding lemma shows that @xmath614 must be of the form @xmath636 for some rational function @xmath637 of a complex variable , and allows us to show that in fact @xmath634 is close to a matrix of the form @xmath638 when the variance of @xmath639 is small .
this follows from the next result .
[ lem : w(t ) ] assume that @xmath168 , and @xmath640 satisfies the inequality @xmath641 for every rank one matrix @xmath612 and all unit vectors @xmath642 .
then for any @xmath643 in the numerical range @xmath644 , we have @xmath645 .
given two unit ( column ) vectors @xmath642 , consider the necessarily rank one matrix @xmath646 .
the hypothesis implies that @xmath647 we deduce that the numerical range @xmath644 has diameter at most @xmath115 , and therefore there @xmath648 for any @xmath649 .
thus any @xmath649 satisfies the conclusion because the norm of an operator is at most twice its numerical radius ( see ( * ? ? ?
* theorem 1.3 - 1 ) ) .
a further property of eigenvectors of hermitian matrices which are close in norm to each other appears in the analysis of the behaviour of the eigenvectors of our matrix models .
the following lemma appears already in @xcite ; we offer a proof for the reader s convenience .
[ eigenspaces ] let @xmath29 and @xmath650 be hermitian @xmath28 matrices .
assume that @xmath651 are such that @xmath652 , @xmath353 , and neither @xmath29 nor @xmath650 has any eigenvalues in @xmath653\cup [ \beta,\beta+\delta]$ ] .
then , @xmath654 in particular , for any unit vector @xmath655 , @xmath656 consider the rectangle @xmath507 having as corners the complex points @xmath657 and @xmath658 . by assumptions ,
we have @xmath659\cup [ \beta,\beta+\delta])=\varnothing$ ] and @xmath660\cup [ \beta,\beta+\delta])=\varnothing$ ] . thus the spectral projections can be obtained by analytic functional calculus : @xmath661\,d\lambda.\ ] ] an application of the resolvent equation and elementary norm estimates yield @xmath662 the lemma follows . for the following concentration of measure
result it is convenient to identify @xmath663 with the subspace of @xmath664 consisting of all vectors whose last component is zero .
similarly , @xmath665 is identified with those matrices in @xmath666 whose last column and row are zero .
we use the notation @xmath667 for variance .
[ concentration ] fix a positive integer @xmath23 , a projection @xmath474 of rank @xmath23 , and a scalar @xmath668 .
then : 1 . @xmath669\,p^*)=0 $ ] almost surely .
given unit vectors @xmath642 , @xmath670 $ ] . assertion ( i ) is equivalent to the statement that , given unit vectors @xmath642 @xmath671)h=0\label{eq:55}\ ] ] almost surely .
the random variable @xmath672 is a lipschitz function on the unitary group @xmath673 with lipschitz constant @xmath674 .
an application of ( * ? ? ?
* corollary 4.4.28 ) yields the inequality @xmath675\right)h\vert>\frac{\varepsilon}{n^{\frac{1}{2}-\alpha}}\right)\leq2\exp\left(-cn^{2\alpha}\vert\im z\vert^{4}\varepsilon^{2}\right).\ ] ] for any @xmath676 , and follows by an application of the borel - cantelli lemma . to prove ( ii ) , apply the same inequality in the usual formula @xmath677=\int_{0}^{+\infty}\mathbb{p}(x > t)\ , dt$ ] for a positive random variable @xmath29 . in the following result
the coefficient @xmath678 can be replaced by @xmath679 if we estimate the operator norm of a matrix by its hilbert - schmidt norm .
[ cor : variance - estimate]fix a positive integer @xmath680 , matrices @xmath681 of rank at most @xmath680 , and a scalar @xmath668 .
then : @xmath682)z\|^{2}]\le ct^{4}\|y\|^{2}\|z\|^{2}/[n|\im z|^{4}].\ ] ] choose orthonormal vectors @xmath683 whose span contains the range of @xmath604 and orthonormal vectors @xmath684 whose span contains the range of @xmath685 .
the corollary follows from the inequality @xmath686)z\|\le\sum_{i , j=1}^{t}\|y\|\|z\|\vert k_{j}^{*}\left(r_{n}(z)-\mathbb{e}[r_{n}(z)]\right)h_{i}\vert\ ] ] and part ( ii ) of the preceding lemma . [ 4.13 ] we note for further use that lemma [ concentration ] and corollary [ cor : variance - estimate ] apply to the resolvent of any selfadjoint polynomial in @xmath687 noncommuting variables @xmath688 as long as the norms of @xmath689 and @xmath18 are uniformly bounded in @xmath2 .
the three subsections below provide a parallel treatment of the three models under consideration .
we use the notation from subsection [ subsec:+perturb ] . fix @xmath690 and @xmath691 . due to the left and right invariance of the haar measure on @xmath5 we may , and do , assume without loss of generality that both @xmath0 and @xmath1 are diagonal matrices .
more precisely , we let @xmath17 be the diagonal matrix @xmath692 where @xmath693 .
we also have @xmath694 , but no order relation is assumed between @xmath695 and @xmath696 other than @xmath697 . for @xmath142 , we write @xmath698 where @xmath699 and @xmath700 we have @xmath701 , where @xmath162 is the @xmath702 matrix representing the usual projection @xmath703 onto the first @xmath137 coordinates , and @xmath704 the operator @xmath162 is precisely @xmath705 corestricted to its range .
similarly , @xmath706 where @xmath707 [ [ reduction - to - the - almost - sure - convergence - of - a - ptimes - p - matrix551 ] ] reduction to the almost sure convergence of a @xmath712 matrix[551 ] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ here we explain how to reduce , in the spirit of @xcite , the problem of locating outliers of @xmath713 to a convergence problem for a random matrix of fixed size @xmath712 .
the matrices @xmath714 and @xmath715 have no spikes , and therefore ( * ? ? ?
* corollary 3.1 ) applies to the matrix @xmath716 .
recall that @xmath717 .
we reformulate this result as follows : there exist positive random variables @xmath718 , such that @xmath719 and @xmath720 almost surely . given @xmath721
, we have @xmath722 and therefore @xmath723 using the fact that @xmath724 when @xmath725 and @xmath726 are square , we obtain @xmath727 so @xmath728 we conclude that the eigenvalues of @xmath713 outside @xmath729 are precisely the zeros of the function @xmath730 , where @xmath731 in that open set .
this is a random analytic function defined on @xmath732 , with values in @xmath733 .
we argue next that the sequence @xmath734 converges almost surely to the deterministic diagonal matrix function @xmath735 where @xmath129 is the subordination function from ( [ subord1 ] ) .
[ estimationinmean ] fix a positive integer @xmath137 , and let @xmath482 and @xmath736 be deterministic real diagonal @xmath737 matrices whose norms are uniformly bounded and such that the limits @xmath738 exist for @xmath739 suppose that the empirical eigenvalue distributions of @xmath482 and @xmath736 converge weakly to @xmath6 and @xmath7 , respectively
. then the resolvent @xmath740 satisfies @xmath741p_n^*=\mathrm{diag}\left ( \frac{1}{\omega_1(z)-\displaystyle\eta_{1}},\ldots , \frac{1}{\displaystyle\omega_1(z)-\eta_{p}}\right),\label{secondapproximation}\ ] ] where @xmath129 is the subordination function from .
since all functions involved satisfy @xmath742 , it suffices to consider the case of @xmath249 .
fix such a scalar @xmath31 and apply lemma [ bicommutant](2 ) to @xmath743 to conclude that the @xmath28 matrix @xmath744 $ ] is diagonal . set @xmath745_{ii}}+(c_n)_{ii},\quad 1\leq i\leq p.\ ] ] and observe that @xmath746 for @xmath747 by lemma [ l ] .
we proceed to show that this function satisfies an approximate subordination relation .
we state this separately for future reference .
the existence of @xmath749^{-1}+c_n,\quad z\in\mathbb{c}^{+},\ ] ] is guaranteed by lemma [ l ] .
we apply now lemma [ bicommutant ] with @xmath750 in place of @xmath38 and @xmath736 in place of @xmath18 , so @xmath744=g(zi_n - c_n)$ ] .
relation ( [ eq : second - commutation ] ) shows that @xmath751^{-1 } \mathbb{e}\big[({r}_{n}(z)-\mathbb{e}[{r}_{n}(z ) ] ) } \\ & \quad & \times(yc_n - c_ny)({r}_{n}(z)-\mathbb{e}[{r}_{n}(z)])\big]\mathbb{e}[{r}_{n}(z)]^{-1},\quad y\in m_{n}(\mathbb{c}).\end{aligned}\ ] ] suppose that @xmath752 has rank one and @xmath642 are unit vectors . in this case , there exist rank one projections @xmath753 and rank two projections @xmath754 ( depending on @xmath31 ) such that @xmath755^{-1}}\\ & & \mbox{}\times\mathbb{e}[p_{1}({r}_{n}(z)-\mathbb{e}[{r}_{n}(z)])q_{1}(yc_n - c_ny)q_2({r}_{n } ( z)-\mathbb{e}[{r}_{n}(z)])p_{2}]\\ & & \mbox{}\times\mathbb{e}[{r}_{n}(z)]^{-1}h.\end{aligned}\ ] ] indeed , the third and first factors in the product above have rank one , while the rank of @xmath756 is at most two .
we deduce that @xmath757)q_{1}\|^{2}]^{1/2}\\ & & \times \mathbb{e}[\|q_{2}({r}_{n}(z)-\mathbb{e}[{r}_{n}(z)])p_{2}\|^{2}]^{1/2}.\end{aligned}\ ] ] we use now the estimates from lemma [ l ] and corollary [ cor : variance - estimate ] along with the inequality @xmath758 to obtain a constant @xmath500 ( independent of @xmath2 and @xmath31 ) such that @xmath759 the number @xmath760 is precisely the @xmath761 entry of the matrix @xmath762 and thus it belongs to the numerical range @xmath763 ; indeed it equals @xmath764 , where @xmath765 is the basis in which @xmath482 is diagonal . lemma [ lem : w(t ) ] yields the estimate @xmath766 which gives the desired result as @xmath77 .
fix now @xmath767 and observe that the family of functions @xmath768 is normal on @xmath769 .
voiculescu s asymptotic freeness result shows that @xmath770)=g_{\mu\boxplus\nu}(z ) = g_\nu(\omega_1(z)),\ ] ] so lemma [ approximatesubordination ] implies that @xmath771 converges uniformly on compact subsets of @xmath769 .
this , together with a second application of lemma [ approximatesubordination ] , implies and concludes the proof of proposition [ estimationinmean ] . the convergence result for the functions @xmath772 also uses a normal family argument ,
more specifically the fact that a normal sequence converges uniformly on compact sets if it converges pointwise on a set with an accumulation point which belongs to the domain .
lemma [ lem : extension ] , part ( b ) , and the hypothesis on @xmath255 , show that the function @xmath776 is analytic on @xmath777 .
define @xmath778 the first @xmath137 diagonal elements of @xmath714 are all equal to @xmath255 and thus lemma [ concentration](i ) and ( applied to @xmath779 and @xmath780 ) show that given @xmath781 , the sequence @xmath782 converges almost surely to @xmath783 moreover , by @xcite , these functions are almost surely uniformly bounded on any compact subset of @xmath774 .
uniform boundedness on some neighbourhood of infinity in @xmath784 is automatic .
we deduce that , almost surely , this sequence converges uniformly on compact subsets of @xmath546 to the function @xmath785 .
the proposition follows immediately from these facts and .
[ proof of theorem [ main+ ] , parts ( 1 ) and ( 2)eigenvalue behaviour .
] we proceed in two steps . * step 1 . *
we investigate first the case in which @xmath786 , that is , @xmath787 has no spikes .
equivalently , we prove our result for the simpler model @xmath788 .
we can work on the almost sure event , whose existence is guaranteed by proposition [ uniformconvergence ] , on which we apply lemma [ alt - benaych - rao ] on this event , with @xmath556 .
we first argue that the hypotheses of that lemma are satisfied .
the values of @xmath531 are clearly diagonal matrices and @xmath791 . we show that the zeros of @xmath792 are simple .
indeed , @xmath793 and the zeros of @xmath794 are simple by the julia - carathodory theorem ( * ? ? ?
* chapter i , exercises 6 and 7 ) because @xmath795 , hypotheses ( 1 ) and ( 3 ) of lemma [ alt - benaych - rao ] follow from proposition [ uniformconvergence ] . to verify hypothesis ( 2 ) of lemma [ alt - benaych - rao ] , observe that if @xmath796 is not invertible then @xmath31 is an eigenvalue of the selfadjoint matrix @xmath797 , hence real .
there are arbitrarily small numbers @xmath353 such that the boundary points of @xmath798 are not zeroes of @xmath520 .
when this condition is satisfied , lemma [ alt - benaych - rao ] yields precisely the conclusion of theorem [ main+](1)(2 ) , when @xmath786 .
indeed , as explained in subsection [ 551 ] , the eigenvalues of @xmath713 in @xmath799 are exactly the zeroes of @xmath539 , and the set of points @xmath31 such that @xmath514 is not invertible is precisely @xmath800 .
this completes the first step . *
* suppose now that @xmath801 and use step 1 above to obtain the existence of a sequence of positive random variables @xmath802 such that @xmath720 almost surely and @xmath803 where @xmath804 we proceed as in step 1 ( switching the roles of @xmath17 and @xmath18 ) in order to conclude that the eigenvalues of @xmath62 outside @xmath805 are precisely the zeros of the function @xmath806 in that open set , where @xmath807 and @xmath163 is the orthogonal projection @xmath711 .
lemma [ alt - benaych - rao ] is applied now to the functions @xmath808 @xmath809 and the compact set @xmath810 .
the convergence of @xmath811 to @xmath812 follows by an adaptation of proposition [ uniformconvergence ] .
this completes the proof of parts ( 1 ) and ( 2 ) of theorem [ main+ ] in the general case @xmath801 , provided that @xmath180 . by symmetry
, we have also proved these assertions in case @xmath178 . to prove part ( 2 ) in case @xmath813 , we use a perturbation argument .
fix @xmath814 such that @xmath815 for some @xmath816 and @xmath817 for some @xmath818 , and fix @xmath168 as in the statement of ( 2 ) . choose @xmath819 so small that @xmath820 contains no spikes @xmath821 and @xmath822 contains no spikes @xmath823 .
since @xmath824 is strictly increasing on @xmath825 , we have @xmath826 , with @xmath827 .
we use the already established part ( 2 ) of the theorem to conclude that , almost surely for large @xmath2 , the perturbed matrix @xmath828 has @xmath829 eigenvalues in @xmath510 and another @xmath830 eigenvalues in the disjoint interval @xmath831 .
an application of lemma [ eigenspaces ] for sufficiently small @xmath508 shows that @xmath62 has @xmath832 eigenvalues in @xmath220 .
[ proof of theorem [ main+ ] , parts ( 3 ) and ( 4)eigenspace behaviour . ]
we borrow heavily from the techniques of @xcite .
there are again two steps .
* step a. * we prove parts ( 3 ) and ( 4 ) of theorem [ main+ ] under the additional assumption that @xmath833 , @xmath834 , @xmath835 , and @xmath178 .
thus @xmath836 for some @xmath816 , and @xmath837 .
assertion ( 3 ) of theorem [ main+ ] follows if the equalities @xmath838 and @xmath839 are shown to hold almost surely for all @xmath840 the hermitian matrices in these equations have rank one , so their norm is equal to the absolute value of their trace . thus we need to show that @xmath841=\frac{\delta_{i_0i}}{\omega_1'(\rho)},\quad i=1,\dots , p,\ ] ] and @xmath842=0,\quad j=1,\dots , q,\ ] ] almost surely .
it is useful to write the random variable in in terms of functional calculus with _ continuous _ rather than indicator functions .
choose @xmath353 so small that each interval @xmath843 $ ] contains exactly one point of @xmath844 ( namely , @xmath695 ) for @xmath845 and for large @xmath2 .
for each @xmath845 , choose a function @xmath846 with support in @xmath843 $ ] such that @xmath847 and @xmath848 . also choose a function @xmath849 with support in @xmath850 $ ] such that @xmath851 and @xmath852 for @xmath853 $ ] .
for sufficiently large @xmath2 , we have @xmath854 . also , by the already established assertion ( 1 ) of the theorem
, we have @xmath855 almost surely for @xmath2 sufficiently large .
thus we see that , almost surely for sufficiently large @xmath2 , @xmath856= { \rm tr}_n\left[h(x_n ) f_i(a_n)\right],\quad i=1,\dots , p.\ ] ] to conclude the proof of , we obtain as in lemma [ concentration ] a concentration inequality for the right - hand side of and then we estimate the expected value . in the following argument
we use the fact that a lipschitz function on @xmath74 remains lipschitz , with the same constant , when considered as a function on the selfadjoint matrices endowed with the hilbert - schmidt norm . see ( * ? ? ?
* lemma a.2 ) for a simple proof of this fact , first observed in @xcite .
[ inegconc]fix @xmath857 , denote by @xmath507 the lipschitz constant of the function @xmath858 , and set @xmath859 . for @xmath2
sufficiently large , the random variable @xmath860 $ ] satisfies the concentration inequality @xmath861 the lemma follows from ( * ? ? ?
* corollary 4.4.28 ) once we establish that the lipschitz constant of the function @xmath862,\quad u\in{\rm u}(n),\ ] ] is at most @xmath863 . for any @xmath405 and @xmath63 in @xmath5 we have @xmath864|\\ & \leq & \|h ( a_n+u^*b_nu ) -h ( a_n+v^*b_nv)\|_2\\ & \leq&\gamma \vert u^ * b_n u -v^ * b_n v \vert_2 , \end{aligned}\ ] ] where we used the cauchy - schwarz inequality for the hilbert - schmidt norm and the fact that @xmath865 . since @xmath866
we conclude that @xmath867 as desired . the above result , combined with the borel - cantelli lemma , yields immediately @xmath868-\mathbb{e}\left [ { \rm tr}_n \left [ h ( x_n ) f_i(a_n)\right]\right]\right)=0,\quad i=1,\dots , p,\ ] ] almost surely .
we conclude the proof of by showing that @xmath869\right]=\frac{\delta_{i_0i}}{\omega'_1(\rho ) } , \quad i=1,\dots , p.\ ] ] lemma [ approxcauchy ] with @xmath870 allows us to rewrite this as @xmath871 \right]h(t)\,dt=- \frac{\delta_{i_0i}}{\omega'_1(\rho)},\quad i=1,\dots , p,\ ] ] or more simply , because @xmath872 is the projection of @xmath663 onto the @xmath873th coordinate , @xmath874_{ii } \right]h(t)\,dt=- \frac{\delta_{i_0i}}{\omega'_1(\rho)},\quad i=1,\dots , p.\ ] ] lemma [ approximatesubordination ] suggests writing @xmath875 = \frac{1}{\omega_1(z)-\theta_i } + \delta_{i , n}(z),\quad i=1,\dots , p , z\in\mathbb{c}^+.\ ] ] we proceed to estimate the functions @xmath876 .
define analytic functions @xmath771 for @xmath845 using with @xmath880 in place of @xmath881 , respectively .
these functions are analytic outside the interval @xmath882 $ ] , and the hypothesis that @xmath883 and @xmath884 are uniformly bounded implies their analyticity on @xmath885 $ ] for some @xmath558 independent of @xmath2 .
the operator @xmath886 $ ] belongs to @xmath887 by lemma [ bicommutant ] , and is therefore a diagonal operator , so @xmath888^{-1}\right)_{ii } + \theta_i,\quad i=1,\dots , p.\ ] ] by lemma [ l ] and the considerations preceding it ( especially and ) , @xmath889)_{ii}+\theta_i=-{\rm tr}_n(b_n),\end{aligned}\ ] ] and @xmath890)^2]_{ii}\\ & = & -(\mathrm{tr}_n(b_n^2)-\mathrm{tr}_n(b_n)^2),\end{aligned}\ ] ] for @xmath142 and @xmath845 .
it follows that we have , as in , nevanlinna representations of the form @xmath891}\frac{1}{z - t}\,d\sigma_{n , i}(t),\quad z\in\mathbb c\setminus[-m , m],\ ] ] where @xmath892 is a positive measure on @xmath893 $ ] , with total mass @xmath894 .
similarly , the subordination function @xmath129 from can be written as @xmath895 the hypothesis that the empirical eigenvalue distribution of @xmath18 converges to @xmath7 implies in particular @xmath896 in addition , the fact that @xmath897 uniformly on compact subsets of @xmath898 $ ] implies that @xmath899 is supported in @xmath893 $ ] and that @xmath900 in the weak@xmath437-topology .
we also have @xmath901 lemma [ conv ] , applied to the sequence @xmath902 yields positive numbers @xmath903 such that @xmath904 and @xmath905 we can now estimate @xmath906-\frac{1}{\omega_1(z)-\theta_i}\right|&= & \left|\frac{1}{\omega_{n , i}(z)-\theta_i}-\frac{1}{\omega_1(z)-\theta_i}\right|\\ & = & \frac{|\omega_{n , i}(z)-\omega_1(z)|}{|\omega_{n , i}(z)-\theta_i||\omega_{1}(z)-\theta_i|}\\ & < & \frac{|\omega_{n , i}(z)-\omega_1(z)|}{|\im z|^2}\\ & \leq&a_n(1+|z|)^4\cdot(1+|\im z|^{-4}),\end{aligned}\ ] ] where @xmath907 and @xmath908 the proposition follows .
the preceding result , combined with , shows that is equivalent to @xmath912 this is easily verified
. indeed , denote by @xmath913 , @xmath247 , the rectangle with vertices @xmath914 .
calculus of residues yields @xmath915 on the other hand , @xmath916\,dt.\ ] ] now we use the fact that @xmath917 on @xmath918 to conclude that @xmath919 is a sum of the following four integrals : @xmath920 @xmath921 all of which are easily seen to tend to zero as @xmath922 .
this completes the proof of and therefore of .
we observe now that the proof of for @xmath923 uses only the fact that @xmath924 . therefore switching the roles of @xmath17 and @xmath18 in this argument
yields a proof of and completes the proof of part ( 3 ) of theorem [ main+ ] in this case if @xmath835 and @xmath178 . the case @xmath925 and @xmath180 follows by symmetry . assertion ( 4 ) of the theorem follows from ( 3 ) simply because @xmath926 is a projection of rank one
. indeed , denote by @xmath927 the canonical basis in @xmath928 , so @xmath929 .
let @xmath930 be a unit vector in the range of @xmath185 , so @xmath931 for every @xmath932 .
direct calculation shows that @xmath933 and thus , almost surely for large @xmath2 , @xmath934 in particular , we obtain @xmath935 , which is precisely the first relation in ( 4 ) .
the case @xmath936 follows by symmetry .
* step b : * in this step we prove ( 3 ) and ( 4 ) in the general case of spikes with higher multiplicities and arbitrary values for @xmath829 and @xmath830 .
we use an idea from @xcite to reduce the problem to the case considered in step a. given positive numbers @xmath126 and @xmath508 , set @xmath937 for @xmath938 .
these matrices have distinct spikes @xmath939 and @xmath940 , respectively .
the fact that @xmath129 is increasing and continuous at @xmath81 implies that , for sufficiently small @xmath126 , there exist exactly @xmath829 indices @xmath941 such that the equations @xmath942 each have a solution @xmath943 , @xmath944 .
similarly , for sufficiently small @xmath508 there exist @xmath830 indices @xmath945 and @xmath830 values @xmath946 such that @xmath947 , @xmath948 .
the numbers @xmath126 and @xmath508 can be chosen such that the intervals @xmath949 @xmath950 are pairwise disjoint and contained in @xmath220 .
we conclude that the arguments of step a hold with @xmath951 , @xmath952 , and @xmath126 in place of @xmath62 , @xmath81 , and @xmath115 , respectively .
thus , @xmath953 almost surely for @xmath954 .
we have @xmath955 and also , noting that @xmath956 for @xmath957 , @xmath958 for small @xmath126 .
in addition , @xmath959 can be made arbitrarily close to @xmath960 by making @xmath126 sufficiently small .
we conclude that @xmath961 almost surely for large @xmath2 if @xmath126 is sufficiently small .
clearly , @xmath962 an application of lemma [ eigenspaces ] shows that @xmath963 can be made arbitrarily small for appropriate choices of @xmath126 and @xmath508 , uniformly in @xmath2 .
the first inequality in ( 3 ) follows at once , and the second one is proved similarly .
we now verify assertion ( 4 ) when @xmath178 .
let @xmath930 be a unit vector in the range of @xmath964 .
since the quantity in is small , we can find , almost surely for large @xmath2 , unit vectors @xmath965 such that @xmath966 , uniformly in @xmath2 .
it suffices therefore to prove that @xmath967 can be made arbitrarily small for appropriate choices of @xmath126 and @xmath508 .
write @xmath968 with @xmath969 in the range of @xmath970 , @xmath971 .
the case of assertion ( 4 ) proved in step a shows that for any @xmath972 sufficiently small , @xmath973 we also have @xmath974 for @xmath975 .
since @xmath976 the relation @xmath977 implies @xmath978 the desired conclusion follows by noting that @xmath979 can be made arbitrarily close to @xmath980 .
the second part of assertion ( 4 ) follows by symmetry .
the proofs of the versions of theorem [ mainx+ ] for the positive line and for the unit circle follow the same outline .
we avoid excessive repetition and only indicate the differences in the tools used throughout the proof .
we use the notations from subsection [ subsec : xperturb+ ] . as in the previous section
, we assume that both @xmath17 and @xmath18 are diagonal matrices : @xmath981 @xmath982 since @xmath983 , fix @xmath984 and @xmath985 .
we use the following multiplicative decompositions : @xmath986 @xmath987 @xmath988 @xmath989 @xmath990 @xmath991 [ [ reduction - to - the - almost - sure - convergence - of - a - ptimes - p - matrix521 ] ] reduction to the almost sure convergence of a @xmath712 matrix[521 ] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ we argue first that , in case @xmath999 , it follows that @xmath62 is almost surely invertible for large @xmath2 . indeed , in this case , @xmath1000 .
therefore , @xmath17 and @xmath18 are invertible for large @xmath2 , and thus so is @xmath62 .
this observation allows us to restrict the analysis to nonzero eigenvalues of @xmath62 .
denote @xmath1001 fix @xmath1002 so that the matrix @xmath1003 is invertible . using sylvester s identity @xmath1004
, we obtain for large @xmath2 @xmath1005 the matrix @xmath1006 is of the form : @xmath1007,\ ] ] where @xmath772 is the analytic function with values in @xmath1008 defined on @xmath1009 by @xmath1010 and @xmath1011 is the diagonal @xmath712 matrix defined earlier .
thus , for large @xmath2 , the nonzero eigenvalues of @xmath1012 outside @xmath1013 are precisely the zeros of @xmath1014 in that open set . as in subsection
[ plus ] , the random matrix functions sequence @xmath773 converges a.s . to a diagonal deterministic @xmath712
matrix function : [ uniformconvergence ] fix a positive integer @xmath137 , and let @xmath482 and @xmath736 be deterministic nonnegative diagonal @xmath28 matrices with uniformly bounded norms such that the limits @xmath738 exist for @xmath1015 .
suppose that the empirical eigenvalue distributions of @xmath482 and @xmath736 converge weakly to @xmath6 and @xmath7 , respectively .
then the resolvent @xmath1016 satisfies @xmath1017p_n^*=\mathrm{diag}\left(\frac{1}{1-\eta_1 \omega_1(z^{-1})},\ldots,\frac{1}{1-\eta_p \omega_1(z^{-1})}\right).\ ] ] we consider without loss of generality elements @xmath249 .
if @xmath482 is invertible , then lemma [ bicommutant ] , part ( 2 ) , applied to @xmath1018 implies that @xmath1019 $ ] is diagonal .
if @xmath482 is not invertible , then @xmath1020=\mathbb{e } [ { r}_n(z)].\ ] ] the limit of diagonal matrices is diagonal , so @xmath1021 $ ] is diagonal .
define @xmath1022_{ii}}\right),\quad1\leq i\leq p.\ ] ] we prove the uniform convergence on compact subsets of @xmath1023 of the sequences @xmath1024 of analytic functions to @xmath129 .
the multiplicative counterpart of lemma [ approximatesubordination ] is as follows : [ approximatesubordination ] assume that @xmath1025 for some @xmath168 .
we have : @xmath1026-\left(i_n-\omega _ { n , i}\left(z^{-1}\right)c_n\right)^{-1}\right\| = 0,\quad z\in \mathbb{c}\setminus \mathbb{r},i\in\{1,\dots , p\}.\ ] ] for @xmath747 , define @xmath1027^{-1}= \mathbb e\left[\left((zc_n)^{-1}-u_n^*d_nu_n\right)^{-1}\right]^{-1}.\ ] ] this function is well - defined by lemma [ l ] , and the second equality is justified by lemma [ bicommutant](2 ) .
we apply lemma [ bicommutant](1 ) with @xmath1028 to obtain @xmath1029\omega_n(z).\end{aligned}\ ] ] consider arbitrary vectors of norm one @xmath1030 and an @xmath752 of rank one to conclude the existence of rank one projections @xmath1031 and rank 2 projections @xmath1032 so that @xmath1033^{1/2}\\ & & \mbox{}\times\mathbb e\left[\left\|q_2\left(\left((zc_n)^{-1}-u_n^*d_nu_n\right)^{-1}-\omega_n(z)^{-1}\right)p_2\right\|^2\right]^{1/2}.\end{aligned}\ ] ] lemma [ l ] yields @xmath1034 , with @xmath1035
. remark [ 4.13 ] provides estimates for the last two factors .
the estimate @xmath1036 is obvious .
thus , @xmath1037 for some constant @xmath1038 independent of @xmath2 .
the @xmath761 entry of the matrix @xmath1039 is precisely @xmath1040 , which belongs to the numerical range of @xmath1039 .
lemma [ lem : w(t ) ] yields @xmath1041 since @xmath482 is diagonal , the lemma follows by letting @xmath77 .
the proof of proposition [ uniformconvergence ] @xmath482 is bounded from below by a positive multiple of @xmath492 is now completed by an application of the above lemma . indeed , using biane s subordination formula and the asymptotic freeness result of voiculescu @xcite , we obtain @xmath1042)=1+\psi_{\mu\boxtimes\nu } ( 1/z)=1+\psi_\mu(\omega_1(1/z))$ ] . clearly , @xmath1043 .
the result follows by analytic continuation .
the general case follows by replacing a non - invertible @xmath482 by @xmath1044 .
the approximation is uniform in @xmath2 , so a normal family argument yields the desired result as @xmath1045 . observe that @xmath1046 where @xmath1047 denotes the resolvent of @xmath1048 .
an application of proposition [ uniformconvergence ] to @xmath1049 and @xmath1050 , remark [ 4.13 ] , as well as of lemma [ lem : extensionx ] , yield the following result .
we leave the details , similar to the ones in the proof of proposition [ uniformconvergence ] , to the reader .
[ [ proofs - of - the - main - results - for - the - positive - multiplicative - model . ] ] proofs of the main results for the positive multiplicative model . ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ [ proof of theorem [ mainx+ ] , parts ( 1 ) and ( 2)eigenvalue behaviour . ]
@xmath1053 * step 1 .
* we prove our result for the model @xmath1054 in which only @xmath17 has spikes .
we consider the almost sure event , whose existence is guaranteed by proposition [ estimationinmean ] , on which there exist a random sequence @xmath1055 converging to zero so that : on this event , we apply lemma [ alt - benaych - rao ] with @xmath1060 , the sequence @xmath1061 and its uniform on compacts limit @xmath531 .
we argue first that the function @xmath796 given by equation is invertible for @xmath1062 . indeed ,
the relations preceding imply that , if @xmath796 is not invertible , then @xmath31 is an eigenvalue of the selfadjoint matrix @xmath1012 , and hence it is a real number .
this verifies hypothesis ( 2 ) .
hypotheses ( 1 ) and ( 3 ) follow from proposition [ estimationinmean ] . finally , @xmath1063 , @xmath1064 and the zeros of @xmath794 are simple by the julia - carathodory theorem .
thus , lemma [ alt - benaych - rao ] applies to @xmath772 and @xmath531 . for almost every @xmath353 ,
the boundary points of @xmath798 are not zeroes of @xmath520 .
when this condition is satisfied , lemma [ alt - benaych - rao ] yields precisely the conclusion of theorem [ mainx+ ] ( 1)(2 ) , when @xmath786 . indeed , as noted above , the nonzero eigenvalues of @xmath1012 in @xmath799 are exactly the zeroes of @xmath539 , and the set of points @xmath31 such that @xmath514 is not invertible is precisely @xmath1065 .
this completes the first step . *
* this is completely analogous to the reasoning from the second step of the proof of theorem [ main+](1)(2 ) .
we omit the details .
@xmath1053 * step a. * we assume first that @xmath1066 , @xmath1067 , @xmath178 and @xmath835 .
step a of the proof of theorem [ main+ ] is modified as follows : @xmath62 is replaced by @xmath1068 , the analogue of lemma [ inegconc ] holds with the constant @xmath1069 replaced by @xmath1070 , and proposition [ estimfonda ] is replaced by the following statement .
[ estimfondax ] there is a polynomial @xmath474 with nonnegative coefficients , a sequence @xmath877 of nonnegative real numbers converging to zero when @xmath2 goes to infinity and some nonnegative integer number @xmath680 , such that for every @xmath1015 and @xmath1071 , @xmath1072 = \frac{1}{z } \frac{1}{1-\theta_i\omega_1(1/z ) } + \delta_{i , n}(z),\ ] ] with @xmath1073 we set @xmath1074_{ii}}\right),\quad z\in \mathbb c\setminus[0,+\infty).\ ] ] as established in proposition [ uniformconvergence ] , @xmath1075_{ii}= ( 1-\theta_{i}\omega_1(1/z))^{-1}$ ] .
it follows that @xmath771 converges to @xmath129 uniformly on compacts of @xmath1076 clearly , @xmath771 is also defined on a neighbourhood of zero .
note that @xmath1077_{ii}}\right)= \frac{1}{\theta_i}\left(1-\frac11\right)=0,\ ] ] and @xmath1078_{ii}}{\mathbb e\left[iy(iy+x_n)^{-1}\right]_{ii}}\\ & = & -\frac{1}{\theta_i } \mathbb e[x_n]_{ii}.\end{aligned}\ ] ] in addition , since @xmath1079 which is uniformly bounded , the map @xmath1080 is analytic and real on the complement of an interval @xmath1081 $ ] , with @xmath1082 .
thus , the maps @xmath1080 and @xmath1083 are nevanlinna maps , and hence can pe represented as @xmath1084}\frac{1}{t - z}\,d\phi_{n , i}(t),\quad z\in\mathbb c^+,\ ] ] and @xmath1085}\frac{1}{t - z}\,d\phi(t),\quad z\in\mathbb c^+.\ ] ] here @xmath1086 are positive measures on @xmath1087 $ ] @xmath1088)= \frac{1}{\theta_i}\mathbb e\left[x_n-\chi_{\{0\}}(x_n)\right]_{ii}$ ] , and @xmath1089)=\frac{\int_\mathbb r t\,d(\mu\boxtimes\nu)(t)}{\int_\mathbb r t\,d\mu(t)}=\int_\mathbb rt\,d\nu(t)$ ] .
thus , lemma [ conv ] applies to @xmath1090 to allow the estimate @xmath1091 with @xmath1092)+\sup_n\phi_{n , i}([0,m])$ ] .
we have @xmath1093_{ii}-\frac{1}{z}\cdot\frac{1}{1-\theta_i\omega_1\left(\frac{1}{z}\right)}\right|&=&\left|\frac{1}{z}\cdot\frac{1}{1-\theta_i\omega_{n , i}\left(\frac{1}{z}\right)}-\frac{1}{z}\cdot\frac{1}{1-\theta_i\omega_1\left(\frac{1}{z}\right)}\right|\\ & = & \frac{\theta_i}{|z|}\frac{\left|\omega_{n , i}\left(\frac{1}{z}\right)-\omega_1\left(\frac{1}{z } \right)\right|}{\left|(1-\theta_i\omega_1\left(\frac{1}{z}\right))(1-\theta_l\omega_{n , i}\left(\frac{1}{z}\right))\right|}\\ & < & \frac{1}{|z|}\frac{|z|^2+c_i}{(\im z)^4}\frac{(|z|+m)^4}{\theta_i\phi_{n , i}([0,m ] ) \phi([0,m])}v_{n , i}.\end{aligned}\ ] ] the proposition follows . to complete the argument of step
a , it suffices now to observe that the residue of the function @xmath1094 at @xmath81 is equal to @xmath1095 * step b : * we use the same perturbation argument as in step b of the proof of theorem [ main+ ] .
we reduce the problem to the case of a spike with multiplicity one , to which we apply step a. the only change from the argument in step b of theorem [ main+ ] comes from the form of the subordination functions .
we use perturbations and and define @xmath1096 .
the quantity @xmath1097 tends to zero uniformly in @xmath2 as @xmath1098 .
the details are omitted .
we use the notation from subsection [ subsec : xperturbt ] .
the tools used are identical to the ones used in the analysis of the positive model @xmath191 . however , the domains of definition of the analytic transforms involved are different .
we indicate the relevant differences .
choose @xmath1099 such that @xmath1100 and @xmath1101 .
the reduction to the almost sure convergence of a @xmath712 matrix is performed the same way , and the same concentration inequality holds ( this time with lipschitz constant @xmath1102 ) in lemma [ concentration ] .
the counterparts of propositions [ uniformconvergence ] and [ estimationinmean ] hold , but in proposition [ estimationinmean ] we must consider @xmath1103 .
the resolvent @xmath490 is defined by @xmath1104 .
the function @xmath771 defined by @xmath1105_{ii}}\right ) , \quad z\in\mathbb d,\ ] ] is easily seen to map @xmath318 into itself and fix the origin .
indeed , @xmath1106 . in the unitary version of lemma [ approximatesubordination ] , no supplementary condition on @xmath1107 is required , and for @xmath1108 defined as in the proof of lemma [ approximatesubordination ] , the estimate becomes @xmath1109 if @xmath314 . the estimates for the corresponding resolvents
are provided by lemma [ l ] .
[ proof of theorem [ mainx+ ] for @xmath21 , parts ( 1 ) and ( 2)eigenvalue behaviour . ]
we must now apply lemma [ alt - benaych - rao ] with @xmath557 .
it will be applied to @xmath557 , the sequence @xmath1110 defined by @xmath1111 and the limit @xmath531 provided by proposition [ estimationinmean ] .
observe that @xmath796 is invertible for @xmath1112 . indeed , it is easy to see that , if @xmath796 is not invertible , @xmath31 belongs to the spectrum of the unitary operator @xmath1113 .
the convergence of @xmath772 to @xmath531 follows from the appropriate version of proposition [ estimationinmean ] . clearly , @xmath514 is diagonal and , again by the julia - carathodory theorem , this time applied to the disk , its diagonal entries have only simple zeros .
the remainder of the argument requires no further adjustments .
the relevant changes for this part of the proof occur in proposition [ estimfondax ] , where @xmath1114 must be used instead of @xmath1115 and an application of lemma [ convt ] in place of lemma [ conv ] .
also , the perturbations and are applied to the arguments of @xmath695 and @xmath1116 , respectively .
it is easy to see that our results hold equally well when @xmath17 is random , independent of @xmath56 , and has spikes @xmath1117 with the property that @xmath1118 , @xmath1119 , almost surely .
similarly , @xmath18 can be taken to be random , independent of @xmath17 and @xmath56 , and with spikes @xmath1120 that converge almost surely to @xmath1121 .
the proofs use the general form of propositions [ estimationinmean ] and [ uniformconvergence ] , respectively .
the above remark allows us to treat sums or products of more than two spiked matrices .
more precisely , let @xmath1122 be an integer , let @xmath1123 be deterministic hermitian matrices and let @xmath1124 be independent haar - distributed random matrices .
suppose that the eigenvalue distribution of @xmath689 tends weakly to @xmath1125 and @xmath689 has spikes subject to the hypotheses of subsection [ subsec:+perturb ] .
then @xmath1126 has asymptotic eigenvalue distribution equal to @xmath1127 , and the outliers in the spectrum of @xmath62 are described by an appropriate reformulation of theorem [ main+ ] .
the result can be proved by induction on @xmath829 if we observe that @xmath1128 has the same asymptotic eigenvalue distribution as @xmath1129 , where @xmath1130 and @xmath56 is a haar - distributed unitary independent from @xmath1131 .
a similar remark applies to theorem [ mainx+ ] in the case of the circle . for the case of the multiplicative model on @xmath14
, the corresponding generalization applies to a model of the form @xmath1132 .
[ atoms ] the analogue of theorem [ main+ ] when @xmath1133 was proved in @xcite under the additional assumption that all eigenvalues of @xmath480 except for the spikes are equal to zero .
our arguments provide a proof of this result without this additional assumption .
similar observations apply to theorem [ mainx+ ] when either @xmath6 or @xmath7 is a point mass .
the only case in which one needs to be more careful is that of theorem [ mainx+ ] for the positive half - line when @xmath6 or @xmath7 is equal to @xmath1134 .
suppose , for instance , that @xmath1135 .
the eigenvalues of @xmath1136 are uniformly approximated arbitrarily well by the eigenvalues of @xmath1137 and our methods do apply to the perturbed model @xmath1138 , whose asymptotic eigenvalue distribution is @xmath1139 .
the spikes are calculated explicitly by noting that @xmath1140 , so @xmath1141 , @xmath1142 .
thus , @xmath1143 and @xmath1144 .
the spikes of @xmath1138 are the solutions of the equations @xmath1145 @xmath845 and @xmath1146 , @xmath1147 .
the first set of equations yields the outliers @xmath1148 , @xmath845 , while the second set of equations can be rewritten as @xmath1149 , \quad j=1,\dots , q.\ ] ] as @xmath583 , we conclude that the spikes of @xmath62 are the numbers @xmath1150 , @xmath1151 , where @xmath1152 is the first moment of @xmath6 .
if @xmath1153 , a similar argument shows that @xmath62 has no outliers at all , that is , @xmath1154 almost surely .
[ ex:5.14 ] the following numerical simulation , due to charles bordenave , illustrates the appearance of two outliers arising from a single spike .
we take @xmath1155 and @xmath1156 , where @xmath1157 , with @xmath137 an orthogonal projection of rank @xmath1158 . the matrix @xmath18 is given by the formula @xmath1159,\ ] ] with @xmath415 being sampled from a standard @xmath1160 gue .
m. capitaine , additive / multiplicative free subordination property and limiting eigenvectors of spiked additive deformations of wigner matrices and spiked sample covariance matrices , _ journal of theoretical probability _ , volume 26 ( 3 ) ( 2013 ) , 595648 .
m. capitaine , c. donati - martin , and d. fral , the largest eigenvalues of finite rank deformation of large wigner matrices : convergence and nonuniversality of the fluctuations , _ ann . probab_.
* 37 * ( 2009 ) , 147 .
m. capitaine , c. donati - martin , d. fral and m. fvrier , free convolution with a semi - circular distribution and eigenvalues of spiked deformations of wigner matrices , _ electronic journal of probability _ * 16 * ( 2011 ) , 17501792 .
p. loubaton and p. vallet , almost sure localization of the eigenvalues in a gaussian information - plus - noise model .
application to the spiked models , _ electron .
j. probab . _
* 16 * ( 2011 ) , no .
70 , 19341959 .
n. r. rao and j. w. silverstein , fundamental limit of sample generalized eigenvalue based detection of signals in noise using relatively few signal - bearing and noise - only samples , _ ieee journal of selected topics in signal processing _ * 4 * ( 2010 ) , 468480 . | in this paper we characterize the possible outliers in the spectrum of large deformed unitarily invariant additive and multiplicative models , as well as the eigenvectors corresponding to them .
we allow both the non - deformed unitarily invariant model and the perturbation matrix to have non - trivial limiting spectral measures and spiked outliers in their spectrum .
we uncover a remarkable new phenomenon : a single spike can generate asymptotically several outliers in the spectrum of the deformed model .
the free subordination functions play a key role in this analysis . ] |
in 1969 , ( @xcite ) veech found examples of skew products over a rotation of the circle that are minimal but not uniquely ergodic .
these were turned into interval exchange transformations in @xcite .
masur and smillie gave a geometric interpretation of these examples ( see for instance @xcite ) which may be described as follows .
let @xmath1 denote the billiard in a @xmath2 rectangle with a horizontal barrier of length @xmath11 based at the midpoint of a vertical side .
there is a standard unfolding procedure which turns billiards in this polygon into flows along parallel lines on a translation surface .
see figure [ fig : unfolding ] .
[ fig : unfolding ] .,title="fig : " ] the associated translation surface in this case is a double cover of a standard flat torus of area one branched over two points @xmath12 and @xmath13 a horizontal distance @xmath10 apart on the flat torus . see figure [ fig : slit - torus ] .
we denote it by @xmath14 .
[ fig : slit - torus ] .,title="fig : " ] the linear flows on this translation surface preserve lebesgue measure .
what veech showed in these examples is that given @xmath5 with unbounded partial quotients in its continued fraction expansion , there is a @xmath10 such that the flow on @xmath1 in direction with slope @xmath5 is minimal but not uniquely ergodic .
let @xmath4 denote the set of nonergodic directions , i.e. those directions for which lebesgue measure is not ergodic .
it was shown in @xcite that @xmath4 is uncountable if @xmath10 is irrational .
when @xmath10 is rational , a result of veech ( @xcite ) implies that minimal directions are uniquely ergodic ; thus @xmath4 is the set of rational directions and is countable . by a general result of masur ( see @xcite ) , the hausdorff dimension of @xmath4 satisfies @xmath15
. in @xcite cheung proved that this estimate is sharp .
he showed that if @xmath10 is _ diophantine _ , then @xmath16 . recall that @xmath10 is diophantine if there is lower bound of the form @xmath17 controlling how well @xmath10 can be approximated by rationals .
this raises the question of the situation when @xmath10 is irrational but not diophantine ; namely , when @xmath10 is a liouville number .
boshernitzan showed that @xmath18 for a residual ( in particular , uncountable ) set of @xmath10 ( see the appendix in @xcite ) although it is not obvious how to exhibit a specific liouville number in this set . in this paper
, we establish the following dichotomy : [ thm : dichotomy ] let @xmath9 be the sequence of denominators in the continued fraction expansion of @xmath10 .
then @xmath18 or @xmath7 , the latter case occurring if and only if @xmath10 is irrational and @xmath19 we briefly outline the proof of theorem [ thm : dichotomy ] , which naturally divides into two parts : an upper bound argument giving the dimension @xmath6 result and a lower bound argument giving the dimension @xmath7 result . in [ s : nedir ] we discuss the geometry of the surface @xmath14 associated to @xmath1 ; in particular , the ways it can be decomposed into tori glued together along _
slits_. we call this a partition of the surface
. the main object of study in both parts of the theorem concerns the summability of the areas of the changes of the partitions , expressed in terms ( [ ieq : sumx ] ) of the summability of the cross - product of the vectors of the slits .
the starting point for the proof of hausdorff dimension @xmath6 in the case that @xmath20 is theorem [ thm : ce ] from @xcite .
that theorem asserts that to each nonergodic direction @xmath21 there is an associated sequence of slits @xmath22 and loops @xmath23 whose directions converge to @xmath5 and satisfy the summability condition ( [ ieq : sumx ] ) .
the natural language to describe the manner by which a sequence of vectors is associated to a nonergodic direction is within the framework of @xmath24-expansions .
( see [ s : lvdir ] . ) here , @xmath24 denotes a closed discrete subset of @xmath25 satisfying some mild restrictions and in the case when @xmath24 is the set of primitive vectors in @xmath26 this notion reduces to continued fraction expansions .
we also have the notion of _ liouville direction _ ( relative to @xmath24 ) which intuitively refers to a direction that is extremely well approximated by the directions of vectors in @xmath24 .
under fairly general assumptions , which hold for example if @xmath24 is a set of holonomies of saddle connections on a translation surface , the set of liouville directions has hausdorff dimension zero .
( corollary [ cor : lvdir ] ) the proof of hausdorff dimension @xmath6 then reduces to showing that if @xmath10 satisfies ( [ pm : div ] ) , then every minimal nonergodic direction is liouville with respect to the @xmath24 expansion .
this is stated as lemma [ lem : ne->lv ] . for the proof of lemma [ lem : ne->lv ] the key ingedient is lemma [ lem : min : area ] , which gives a _ lower bound _ on cross - products .
it is based on the fact that @xmath27 will be an extremely good approximation to @xmath28 provided the interval @xmath29 $ ] is large enough and also contains @xmath30 not too close to @xmath31 .
( see lemma [ lem : lc ] . )
this idea is motivated by the elementary fact that for any pair of vectors @xmath32 and @xmath33 where @xmath34 with @xmath35 we have @xmath36 unless @xmath37 are parallel to each other , in which case the cross - product vanishes .
we apply lemma [ lem : min : area ] to the sequence @xmath22 associated by theorem [ thm : ce ] to a minimal nonergodic direction @xmath5 . if one assumes , by contradiction , that @xmath5 is _
not _ liouville with respect to the @xmath24-expansion , then lemma [ lem : min : area ] implies that @xmath38 whenever @xmath39 falls in a large interval @xmath29 $ ] .
moreover , the number of such slits is at least a fixed constant times @xmath40 .
thus the sum of the cross - products would be at least @xmath41 the sum over those @xmath42 for which @xmath29 $ ] is large . since ( [ pm : div ] ) still holds if the sum is restricted to those @xmath42 , the summable cross - products condition ( [ ieq : sumx ] ) would be contradicted .
this will then show that @xmath5 is liouville and we will conclude that @xmath18 . the starting point for the dimension @xmath7 argument is theorem [ thm : sumx ] , which is the specialization of a result from @xcite to the case of @xmath14 that says the summability condition ( [ ieq : sumx ] ) is sufficient to guarantee that the limiting direction of a sequence of slit directions is a nonergodic direction .
one proceeds to construct a cantor set of nonergodic directions arising as a limit of directions of slits on the torus .
aspects of this construction were already carried out in @xcite in the case that @xmath10 is diophantine .
for @xmath44 , let @xmath45 be the set of limiting directions obtained from sequences @xmath22 satisfying @xmath46 .
it was shown in @xcite , under the assumption of diophantine @xmath10 , that one can make the series in ( [ ieq : sumx ] ) be dominated by a geometric series of ratio @xmath47 , and then @xmath48 .
the lower bound @xmath7 then follows by taking the limit as @xmath49 tends to one .
the strategy of bounding cross - products using a geometric series fails if only the weaker diophantine condition ( [ pm : conv ] ) is assumed .
in fact , in the large gaps @xmath50 $ ] , as we have indicated , the cross - product is bounded below by @xmath51 .
so if the gaps are large , ( where the notion of `` large '' is to be made precise later ) then there are many terms with cross - products bounded below by @xmath51 and these terms would eventually become larger than the terms in the geometric series .
this suggests modifying the strategy in @xcite by replacing the geometric series used to dominate the series in ( [ ieq : sumx ] ) with a series whose terms @xmath52 are @xmath53 if @xmath39 lies in a large interval @xmath29 $ ] and are otherwise decreasing like a geometric series of ratio @xmath47 for @xmath54 such that @xmath39 lies between successive large intervals .
the number of slits in @xmath29 $ ] is @xmath55 so that @xmath56 restricted to those @xmath54 for which @xmath39 lies in a large interval @xmath29 $ ] is bounded using the assumption ( [ pm : conv ] ) .
the sum of the remaining terms is bounded by the sum of a geometric series times @xmath57 .
this latter sum is finite .
the finiteness then of @xmath56 and therefore ( [ ieq : sumx ] ) ensures that the resulting set @xmath58 .
following @xcite , we seek to build a tree of slits so that by associating intervals about the direction of each slit in the tree , we can give @xmath45 the structure of a cantor set to which standard techniques can be used to give lower estimates on hausdorff dimension .
these techniques require certain `` local estimates '' ( expressed in terms of lower bounds on the number of subintervals and the size of gaps between them ) hold at each stage of the construction . in [ s : cantor ] , we express these local estimates in terms of the parameters @xmath49 and @xmath52 . for slits @xmath59 whose lengths lie in a `` small '' interval @xmath29 $ ] we repeat the construction given in @xcite to construct `` children '' slits from `` parent '' slits .
this is carried out in
[ s : diophantine ] .
in the current situation we have to combine that construction with a new one to deal with slits lengths that lie between consecutive @xmath60 with large ratio .
we call this the `` liouville '' part of @xmath10 .
the construction of new slits from old ones in that case is carried out in [ s : liouville ] .
the construction of the tree of slits and the precise definition of the terms @xmath52 are given in
[ s : init ] and [ s : tree ] .
these sections are the most technical part of the paper .
the main task is to ensure that the recursive procedure for constructing the tree of slits can be continued indefinitely while at the same time ensuring the required local estimates are satisfied in the case of our two constructions . finally , in
[ s : lower ] , we verify that the series @xmath56 is convergent and that the lower bound on @xmath61 can be made arbitrarily close to @xmath7 by choosing the parameter @xmath49 sufficiently close to one .
finally we record the following by - product of our investigation .
associated to any translation surface ( or more generally a holomorphic quadratic differential ) is a teichmller geodesic . for each @xmath62 the riemann surface @xmath63 along the geodesic
is found by expanding along horizontal lines by a factor of @xmath64 and contracting along vertical lines by @xmath64 .
it is known ( see @xcite ) that if the vertical foliation of the quadratic differential is nonergodic , then the associated teichmller geodesic is _ divergent _ , i.e. it eventually leaves every compact subset of the stratum .
the converse is however false .
there are divergent geodesics for which the vertical foliation is uniquely ergodic .
in fact , we have [ thm : divergent ] let @xmath65 denote the set of divergent directions in @xmath1 , i.e. directions for which the associated teichmller geodesic leaves every compact subset of the stratum . remains valid if @xmath65 is interpreted as the set of directions that are divergent in the sense described in the previous footnote .
] then @xmath66 or @xmath7 , with the latter case occurring if and only if @xmath10 is irrational .
the authors would like to thank emanuel nipper and the referee for many helpful comments .
in this section , we establish notation , study partitions of the surface associated to @xmath1 , and recall the summable cross - products condition ( [ ieq : sumx ] ) for detecting nonergodic directions .
let @xmath67 denote the standard flat torus with two marked points .
a _ saddle connection _ on @xmath68 is a straight line that starts and ends in @xmath69 without meeting either point in its interior . by a _ slit _
we mean a saddle connection that joins @xmath12 and @xmath13 , while a _ loop _ is a saddle connection that joins either one of these points to itself .
holonomies of saddle connections will always be represented as a pair of real numbers . in particular , @xmath70 where @xmath71 is the horizontal slit joining @xmath12 to @xmath13 .
the set of holonomies of loops is given by @xmath72 since @xmath10 is irrational , the set of holonomies of slits is given by @xmath73 where @xmath74 note that @xmath75 and @xmath76 are disjoint and that @xmath76 is in one - to - one correspondence with the set of oriented slits .
when we speak of `` the slit @xmath59 ... '' we shall always mean the slit whose holonomy is @xmath59 , while @xmath77 specifies that the orientation is meant to be from @xmath12 to @xmath13 .
also , each @xmath78 corresponds to a pair of loops , one based at each branch point .
the pair of cylinders in @xmath68 bounded by these loops will be denoted by @xmath79 .
the core curves of these cylinders also have @xmath80 as their holonomy .
each slit @xmath81 has two lifts in @xmath14 whose union is a simple closed curve .
we say @xmath81 is _ separating _ if this curve separates @xmath82 into a pair of tori interchanged by the involution of the double cover .
we denote the slit tori by @xmath83 where @xmath84 .
( @xcite ) a slit @xmath59 is separating if and only if @xmath85 for some even integers @xmath86 .
the collection of separating slits have holonomies given by @xmath87 where @xmath88 the cross - product formula from vector calculus expresses the area of the parallelogram spanned by @xmath89 and @xmath80 as @xmath90 where @xmath91 denote the standard skew - symmetric bilinear form on @xmath25 , @xmath92 the euclidean norm , and @xmath5 the angle between @xmath89 and @xmath80 .
it will be convenient to introduce the following .
the distance between the directions of @xmath93 , denoted by @xmath94 , will be measured with respect to inverse slope coordinates .
that is , @xmath94 is the absolute value of the difference between the reciprocals of their slopes .
we have the folllowing analog of the cross - product formula @xmath95 where @xmath96 denotes the absolute value of the @xmath97-coordinate . for our purposes , the vectors we consider will always have directions close to some fixed direction and nothing essential is lost if one chooses to think of @xmath98 as the _ length _ of the vector @xmath80 ( or to think of @xmath94 as the angle between the vectors ) for these notions differ by a ratio that is nearly constant .
in fact , the notations @xmath98 and @xmath94 are intended to remind the reader of euclidean lengths and angles , and in the discussions we shall sometimes refer to them as such .
these nonstandard notions are particularly convenient in calculations as they allows us to avoid trivial approximations involving square roots and the sine function that would otherwise be unavoidable had we instead insisted on the euclidean notions . as will become clear later , the benefits of the nonstandard notions will far outweigh the potential risks of confusion . [
lem : disjoint ] let @xmath79 be the cylinders in @xmath68 determined by @xmath78 .
a slit @xmath59 is contained in one of the cylinders @xmath99 if and only if @xmath100 . to prove necessity
, we note that the area of the cylinder containing the slit is @xmath101 , which is @xmath102 since the complement has positive area . for sufficiency ,
let us first rotate the surface so that @xmath80 is horizontal .
if the slit were not contained in one of the cylinders , then the vertical component of @xmath59 is a ( strictly ) positive linear combination of the heights @xmath103 of the rotated cylinders .
however , the vertical component is given by @xmath104 which is absurd .
let @xmath105 and @xmath78 .
we shall say @xmath59 and @xmath106 are `` related by a dehn twist about @xmath80 '' if they are contained in the same cylinder determined by @xmath80 . if both lie in @xmath107 ( or both in @xmath108 )
then their holonomies are related by @xmath109 for some @xmath110 . in this case
, we refer to @xmath111 as the _ order _ of the dehn twist .
[ lem : twist ] let @xmath105 and @xmath78 .
if @xmath112 then @xmath59 and @xmath106 are related by a dehn twist about @xmath80 .
lemma [ lem : disjoint ] implies each of @xmath59 and @xmath106 is contained in one of the cylinders @xmath113 and @xmath114 determined by @xmath80 .
if they belong to different cylinders , then the sum of the areas of the cylinders would be less than one , which is impossible .
hence , @xmath59 and @xmath106 lie in the same cylinder and , therefore , they are related by a dehn twist about @xmath80 .
suppose @xmath115 are a pair of separating slits .
then we may measure the change in the partitions they determine by @xmath116 there is an ambiguity in this definition arising from the fact that we have not tried to distinguish between @xmath117 and @xmath118 .
let us agree to always take the smaller of the two possibilities , which is at most one as their sum represents the area of @xmath14 .
[ lem : area ] if @xmath115 are separating slits related by a dehn twist about @xmath80 then @xmath119 let @xmath120 be the cylinder that contains both slits and let @xmath121 be the order of the dehn twist relating them .
note that @xmath122 is even .
the slits cross each other , each subdividing the other into @xmath122 segments of equal length .
the symmetric difference between the partitions is a finite union of parallelograms bounded by the lifts of @xmath59 and @xmath106 .
there are @xmath122 parallelograms , each having area @xmath123 and since @xmath124 , we have @xmath125 , giving the lemma .
each separating slit determines a partition of @xmath14 into a pair of slit tori of equal area .
the next theorem explains how nonergodic directions arise as certain limits of such partitions .
it is a special case , adapted to branched double covers of tori , of a more general condition developed in @xcite that applies to arbitrary translation surfaces and quadratic differentials .
we will use it in [ s : lower ] to identify large subsets of @xmath4 .
[ thm : sumx ] let @xmath22 be a sequence of separating slits with increasing lengths @xmath39 and suppose that every consecutive pair of slits @xmath126 and @xmath127 are related by a dehn twist about some @xmath128 such that @xmath129 then the inverse slopes of @xmath126 converge to some @xmath5 and this limiting direction belongs to @xmath4 . since @xmath130
, we have @xmath131 so that @xmath132 from which the existence of the limit @xmath5 follows . let @xmath133 be the normalised area measure on @xmath14 and let @xmath134 be the component of @xmath126 orthogonal to @xmath135 . theorem 2.1 in @xcite asserts that @xmath5 is a nonergodic direction if the following conditions hold : 1 .
@xmath136 , 2 .
@xmath137 for some constants @xmath138 , and 3 .
@xmath139 .
since @xmath140 , ( ii ) is clear , while ( iii ) is a consequence of ( [ ieq : sumx ] ) , by lemma [ lem : area ] .
it remains to verify ( i ) , but this follows easily from @xmath141 since then @xmath142 so that @xmath143 , by ( [ ieq : sumx ] ) . the converse to theorem [ thm : sumx ] also holds .
that is , to each nonergodic direction @xmath5 one can associate a sequence of slits @xmath144 whose directions converge to @xmath5 and such that all the hypotheses of theorem [ thm : sumx ] hold .
the definition of this sequence will be explained next .
in this section we introduce @xmath24-expansions and use them to define the notion of a liouville direction relative to a closed discrete subset @xmath145 . under fairly general assumptions on @xmath24 , the set of liouville directions is shown to have hausdorff dimension zero . given an inverse slope @xmath5 and @xmath146 we define @xmath147 which we shall refer to as the `` horizontal component '' of @xmath80 in the direction @xmath5 .
it represents the absolute value of the @xmath148-coordinate of the vector @xmath149 where @xmath150 is the horizontal shear that sends the direction of @xmath5 to the vertical .
let @xmath24 be a closed discrete subset of @xmath25 and @xmath5 an inverse slope .
a @xmath24-_convergent _ of @xmath5 is any vector @xmath151 that minimizes the expression @xmath152 among all vectors @xmath153 with @xmath154 .
recall that @xmath98 is the absolute value of the @xmath97-coordinate .
we call it the _ height _ of @xmath80 .
can be identified with @xmath155 , so that the height of the vector @xmath80 coincides with the height of the rational . ]
thus , @xmath24-convergents are those vectors in @xmath24 that minimize horizontal components among all vectors in @xmath24 of equal of lesser height .
the @xmath24-_expansion _ of @xmath5 is defined to be the sequence of @xmath24-convergents ordered by increasing height .
if two or more @xmath24-convergents have the same height we choose one and ignore the others .
note that by definition the sequence of heights of @xmath24-expansion is _ strictly _ increasing and , as a consequence , the sequence of horizontal components is strictly decreasing
if @xmath156 then @xmath157 must be greater than @xmath157 , for otherwise @xmath158 would not qualify as a @xmath24-convergent . in the case
when @xmath24 is the set of primitive vectors in @xmath26 , i.e. @xmath159 , the notion of a @xmath24-convergent reduces to the notion from continued fraction theory .
that is , @xmath160 is a @xmath24-convergent of @xmath5 if and only if @xmath161 is a convergent of @xmath5 in the usual sense . has fractional part strictly between @xmath7 and @xmath162 : the integer part of @xmath5
is the _ zeroth order _ convergent of @xmath5 in the usual sense , but nevertheless fails to be a @xmath24-convergent.]a generalisation to higher dimensions ( where @xmath24 is the set of primitive vectors in @xmath163 for @xmath164 ) is given in @xcite .
obviously , we should always assume @xmath24 does not contain the origin , for otherwise the zero vector is the only convergent , independent of @xmath5 .
let us also assume that @xmath24 contains some nonzero vector on the @xmath148-axis , for this ensures that the heights of @xmath24-expansions are well - ordered .
indeed if @xmath165 is a @xmath24-convergent , then all @xmath24-convergents lie in an infinite parallel strip of width @xmath166 about the direction of @xmath5 .
since the set of @xmath24-convergents forms a closed discrete subset of this strip , there is no accumulation point .
hence , if there are infinitely many @xmath24-convergents , their heights increase towards infinity .
one last assumption we shall impose is the finiteness of the `` minkowski '' constant : @xmath167 where the supremum is taken over all bounded , @xmath6-symmetric convex regions disjoint from @xmath24 .
any direction which is not the direction of a vector in @xmath24 will be called _ minimal _ ( relative to @xmath24 ) .
[ lem : minkowski ] assume ( [ minkowski ] ) and that @xmath24 contains a non - zero vector on the @xmath148-axis .
then the @xmath24-expansion of a direction with inverse slope @xmath5 is infinite if and only if @xmath5 is minimal .
if the @xmath24-expansion is finite , take the last convergent . if it does not lie in the direction of @xmath5
, then there is an infinite parallel strip containing the origin with one side the direction of @xmath5 containing no points of @xmath24 , but this is ruled out by ( [ minkowski ] ) .
hence , its direction is @xmath5 , so @xmath5 is not minimal .
conversely , if @xmath5 is not minimal , then there is a vector in @xmath24 in the direction of @xmath5 and it is necessarily a convergent and no other convergent can beat it , so it is the last one in the @xmath24-expansion . there is also a first convergent ; it lies on the @xmath148-axis .
let @xmath148 the horizontal component of the first convergent and @xmath97 the height of the last convergent .
the compact region @xmath168 contains all the @xmath24-convergents .
since @xmath24 is closed , it is compact ; by discreteness , it is finite .
note that the @xmath24-expansion are defined for all directions except the horizontal . in the sequel
, we shall always assume the hypotheses of lemma [ lem : minkowski ] remain in force .
if @xmath5 is an inverse slope and @xmath89 a non - horizontal vector then we shall often write @xmath169 for the absolute difference between the directions .
that is , @xmath170 for any vector @xmath80 whose inverse slope is @xmath5 .
similarly , the notation @xmath171 will be used to mean @xmath172 where @xmath173 .
[ thm : angle ] the sequence of @xmath24-convergents of @xmath5 satisfies , as in ( [ ieq : angle ] ) , means @xmath174 for any @xmath59 whose inverse slope is @xmath5 . ]
@xmath175 consider the parallelogram @xmath176 defined by ( [ def : p(x , y ) ] ) where @xmath177 and @xmath178 .
the base is @xmath179 and the height is @xmath180 . by definition of @xmath181 ,
the interior of @xmath182 is disjoint from @xmath24 so that ( [ minkowski ] ) implies @xmath183 giving the right hand inequality in ( [ ieq : angle ] ) .
since @xmath184 we have @xmath185 , giving the left hand inequality in ( [ ieq : angle ] ) .
recall that an irrational number is diophantine iff the sequence of denominators of its convergents satisfies @xmath186 for some @xmath187 .
otherwise , it is liouville .
this motivates our next definition .
we say a minimal direction is _ diophantine _ relative to @xmath24 if its @xmath24-expansion satisfies @xmath188 for some @xmath187 .
otherwise , it is _ liouville _ relative to @xmath24 .
note that we have a trichotomy : every direction is either diophantine , liouville or not minimal , relative to @xmath24 .
we say @xmath24 has _ polynomial growth _ of rate ( at most ) @xmath189 if @xmath190 where @xmath191 denotes the ball of radius @xmath192 about the origin .
[ lem : e_r ] let @xmath193 be the set of ( inverse slopes of ) directions @xmath5 whose @xmath24-expansions satisfy @xmath194 for infinitely many @xmath42 . if @xmath24 has polynomial growth of rate @xmath189 , then @xmath195 it is enough to bound the hausdorff dimension of the set @xmath196 $ ] for some arbitrary but fixed @xmath197
let @xmath198 be the set of @xmath151 that arise as @xmath24-convergents of some direction whose inverse slope lies in @xmath199 $ ] and such that @xmath200 then @xmath198 is contained in some ball of radius @xmath201 where @xmath202 is a constant depending only on @xmath203 .
let @xmath204 be the closed interval of length @xmath205 centered about the inverse slope of @xmath80 .
then theorem [ thm : angle ] implies every @xmath206 is contained in @xmath204 for infinitely many @xmath207 .
for any @xmath208 let @xmath209 then given @xmath210 we can choose @xmath208 large enough so that @xmath211 is an @xmath212-cover of @xmath213 .
since the number of elements in @xmath198 is bounded by @xmath214 for some @xmath215 we have @xmath216 so that the @xmath217-dimensional hausdorff measure is finite for any @xmath218 .
this shows @xmath219 , from which the lemma follows . by @xcite
( see also @xcite , @xcite ) the set of holonomies of saddle connections on any translation surface satisfies a quadratic growth rate .
[ cor : lvdir ] the set of liouville directions relative to the set of holonomies of saddle connections on a translation surface has hausdorff dimension zero .
in this section we assume the denominators of the convergents of @xmath10 satisfy ( [ pm : div ] ) and set @xmath220 ( recall the sets @xmath75 and @xmath221 were defined in [ s : nedir ] . ) we shall need the following characterisation of nonergodic directions in terms of @xmath24-expansions .
[ thm : ce ] ( @xcite ) let @xmath5 be a minimal .
] direction in @xmath1
. then @xmath5 is nonergodic if and only if its @xmath24-expansion is eventually alternating between loops and separating slits @xmath222 and satisfies the summable cross - products condition ( [ ieq : sumx ] ) .
our goal is to show that @xmath18 under the assumption ( [ pm : div ] ) . by corollary [ cor :
lvdir ] , it is enough to show that every minimal nonergodic direction is liouville relative to @xmath24 .
note that the sufficiency in theorem [ thm : ce ] follows from theorem [ thm : sumx ] since the heights of @xmath24-convergents increase and as soon as @xmath223 then @xmath126 and @xmath127 are related by a dehn twist about @xmath128 , by lemma [ lem : twist ] .
the main point of theorem [ thm : ce ] is that the converse also holds .
observe that our main task has been reduced to a question about the set of possible limits for the directions of certain sequences of vectors in @xmath24 .
in the sequel we shall need the following two standard facts from the theory of continued fractions .
[ thm : cf1]([kh , thm . 9 and 13 ] )
the sequence of convergents of a real number @xmath5 satisfies @xmath224 [ thm : cf2]([kh , thm .
19 ] ) if a reduced fraction satisfies @xmath225 then it is a convergent of @xmath5 .
the next lemma shows that convergents of @xmath10 with @xmath226 give rise to convergents of @xmath28 .
[ lem : lc ] let @xmath85 be a slit and @xmath227 a convergent of @xmath10 such that @xmath228 let @xmath229 denote the fraction @xmath27 in lowest terms .
then @xmath229 is a convergent of @xmath28 and its height satisfies @xmath230 .
furthermore , the height @xmath231 of the next convergent of @xmath28 is larger than @xmath232 . using the right hand side of ( [ ieq : cf1 ] ) and ( [ sc : lc ] )
we get @xmath233 which implies that @xmath229 is a convergent of @xmath28 .
clearly , @xmath234 and since @xmath235 , @xmath30 is divisible by @xmath236 so that @xmath237 .
let @xmath231 be the height of the next convergent of @xmath28 .
from the first inequalities in ( [ ieq : cf1 ] ) and in ( [ ieq : lc ] ) we get @xmath238 so that @xmath239 when the conclusion of lemma [ lem : lc ] holds , we refer to @xmath229 ( or the vector @xmath160 ) as the _ liouville convergent _ of @xmath59 indexed by @xmath42 .
( we shall often blur the distinction between the rational @xmath229 and the vector @xmath80 . )
the terminology of liouville convergent is justified by the sequel both in the dimension @xmath6 result and in the dimension @xmath43 result . in the next lemma
we show that if @xmath240 have their lengths in a range defined by the convergents of @xmath10 and are related by a twist about a loop @xmath80 , then if @xmath80 is not the liouville convergent of @xmath59 , the area interchange determined by @xmath115 will be large .
if @xmath80 is the liouville convergent , then the next slit after @xmath106 will not be in the range .
the summability condition on area exchanges will then imply that there can not be too many slit lengths in the liouville part of @xmath10 ( in the range where @xmath241 is large ) .
consequently the lengths of the slits must grow quickly and we can find covers of the nonergodic set that allow us to prove hausdorff dimension @xmath6 using lemma [ lem : e_r ] . in
[ s : liouville ] we will use liouville convergents to build new children slits out of parent slits . [ lem : min : area ] let @xmath115 be slits such that @xmath115 are related by a dehn twist about @xmath78 and @xmath242 .
suppose further that @xmath243 and let @xmath89 be the liouville convergent of @xmath59 indexed by @xmath42 . regarding @xmath89 as a vector , then either 1 .
@xmath244 and @xmath245 or 2 .
@xmath246 and for any @xmath247 satisfying @xmath248 we have @xmath249 .
we have @xmath109 for some nonzero , even integer @xmath122 , so that @xmath250 let @xmath251 be the inverse slope of @xmath106 .
let @xmath160 . then @xmath252 so that @xmath229 is a convergent of @xmath251 , by ( [ ieq : cf2 ] ) .
let @xmath231 be the height of the next convergent of @xmath251 .
then ( [ ieq : cf1 ] ) implies @xmath253 so that @xmath254 the liouville convergent @xmath255 can not have its height @xmath256 because lemma [ lem : lc ] implies the height @xmath257 of the next convergent of @xmath251 is greater than @xmath258 , contradicting the fact that @xmath259 is the height of a convergent of @xmath251 , namely @xmath229 .
thus , @xmath260 . in case ( i ) , @xmath261 so that @xmath262 . since @xmath263 , the inequality ( [ ieq : min : area ] ) follows . in case ( ii )
, we have @xmath264 , as noted earlier . given @xmath247
, we have @xmath265 from which it follows that @xmath266 .
the hausdorff dimension @xmath6 result now follows from [ lem : ne->lv ] assume @xmath267 holds .
then any minimal @xmath21 is liouville relative to @xmath24 .
let @xmath268 be defined by @xmath269 ; in other words , @xmath270 note that since @xmath271 grows exponentially , we have @xmath272 for any @xmath273 .
hence , ( [ pm : div ] ) implies @xmath274 is unbounded ; moreover , the series in ( [ pm : div ] ) diverges even if we restrict to terms with @xmath275 .
let @xmath21 be a minimal direction for the flow .
then it is minimal relative to @xmath24 and by theorem [ thm : ce ] its @xmath24-expansion eventually alternates @xmath276 between ( separating ) slits and loops such that ( [ ieq : sumx ] ) holds .
let @xmath277 be the collection of indices @xmath54 such that @xmath278 for any @xmath279 we wish to prove that conclusion ( i ) of lemma [ lem : min : area ] holds .
suppose by way of contradiction conclusion ( ii ) holds so that @xmath128 is the liouville convergent of @xmath126 indexed by @xmath42 .
setting @xmath280 by conclusion ( ii ) we have @xmath281 , a contradiction .
thus ( i ) holds and therefore @xmath282 .
suppose @xmath5 is diophantine relative to @xmath24 .
then there exists @xmath187 such that @xmath283 for all @xmath54 .
hence , @xmath284 and since @xmath285 we see that the number of @xmath54 such that @xmath39 lies in an interval of the form @xmath286 $ ] is at least @xmath287 .
it follows that the number of elements in @xmath277 is at least @xmath288 provided @xmath289 for some @xmath290 depending only on @xmath187 . since @xmath291 ( as heights of convergents grow exponentially ) we have @xmath292 which contradicts ( [ ieq : sumx ] )
hence , @xmath5 must be liouville relative to @xmath24 , proving the lemma .
we begin the proof of the hausdorff dimension @xmath43 result .
to construct nonergodic directions , we use theorem [ thm : sumx ] . the general idea is as follows .
starting with an initial slit @xmath293 we will construct a tree of slits . at level @xmath54
we will have a collection of slits of approximately the same length .
for each @xmath59 in this collection we wish to construct new slits of level @xmath294 each having small cross - product with @xmath59 .
depending on the relationship of the length of @xmath59 to the continued fraction expansion of @xmath10 , as specified precisely in [ s : init ] , the construction will be one of two types that will be explained in
[ s : liouville ] and [ s : diophantine ] . in this section , we associate to this tree of slits a cantor set .
for each @xmath54 we will define a set @xmath295 which is a disjoint union of intervals .
the directions of each slit of level @xmath54 will lie in some interval in @xmath295 and the intervals at level @xmath54 will be separated by gaps .
the intervals of level @xmath294 will be nested in the intervals of level @xmath54 .
each nonergodic direction corresponds to a nested intersection of these intervals
. we shall assume the tree of slits satisfy certain assumptions , to be verified later in
[ s : tree ] and [ s : lower ] .
these assumptions , expressed in terms of parameters @xmath44 , @xmath296 and @xmath297 , ensure that certain lower bounds on the hausdorff dimension of the cantor set will hold . to establish lower bounds for hausdorff dimension we will use an estimate of falconer @xcite which we explain next .
let @xmath298 where each @xmath295 is a finite disjoint union of closed intervals and @xmath299 for all @xmath54 .
suppose there are sequences @xmath300 and @xmath301 such that each interval of @xmath295 contains at least @xmath302 intervals of @xmath303 and the smallest gap between any two intervals of @xmath303 is at least @xmath304 .
( note that @xmath300 implies there will always be at least one gap . ) then falconer s lower bound estimate is @xmath305 if @xmath306 , as is necessarily the case if the length of the longest interval in @xmath295 tends to zero as @xmath307 , then @xmath308 where @xmath309
our goal is that for each @xmath210 , we make a construction of a cantor set of nonergodic directions so that each @xmath310 will satisfy @xmath311 given @xmath44 and a sequence of positive @xmath313 ( which will measure the area interchange defined by consecutive slits ) , we shall construct a cantor set @xmath314 depending on parameters @xmath302 and @xmath304 that are expressible in terms of @xmath49 and @xmath52 .
it is based on the assumption , verified later , that we can construct a tree of slits .
we start with an initial slit @xmath293 , the unique slit of level @xmath6 .
inductively , given a slit @xmath126 of level @xmath54 we consider slits of the form @xmath315 where @xmath316 is a primitive vector , i.e. @xmath317 , and satisfies @xmath318 we refer to @xmath319 of the above form as a _ child _ of @xmath126 .
it satisfies @xmath320 the main difficulty in the construction is avoiding slits that have no children at all . to ensure that we can avoid such slits
, we shall only use children with `` nice diophantine properties '' when we assemble the slits for the next level .
however , we shall ensure that at each stage , the number of children ( of a parent slit @xmath59 ) used will be at least @xmath321 where @xmath312 is to be determined later .
for @xmath59 a slit , let @xmath322 denote the interval of length @xmath323 centered about the inverse slope of the direction of @xmath59 .
the following lemma allows us to find estimates for the sizes of intervals and the gaps between them .
[ lem : gaps ] assume @xmath324 and @xmath325 .
let @xmath127 be a child of a slit @xmath126 of level @xmath54 .
then * @xmath326 , and * if @xmath327 is another child of @xmath126 , then @xmath328 since the distance between the directions of @xmath126 and @xmath127 is @xmath329 the first conclusion follows from @xmath330 which holds easily by the assumption on @xmath331 .
the distance between the directions of @xmath127 and @xmath128 is @xmath332 if @xmath333 is another child of @xmath126 then @xmath334 so that by the triangle inequality , @xmath335 since @xmath336 .
therefore , @xmath337 since @xmath324 .
let @xmath338 where the union is taken over all slits of level @xmath54 . from ( [ ieq:|w_j+1| ] ) we have @xmath339 so that the number of children given by ( [ enough ] ) is at least @xmath340 while the smallest gap between the associated intervals is at least @xmath341 by lemma [ lem : gaps ] .
now we express @xmath310 , given by ( [ def : d_j ] ) , in terms of @xmath342 and @xmath312 .
we have @xmath343 so that @xmath344 giving @xmath345 now making @xmath310 close to @xmath7 will mean making @xmath49 close to @xmath162 and making the terms @xmath346 and @xmath347 small .
notice that if @xmath312 and @xmath52 are constant sequences , then this is easily accomplished by choosing @xmath331 large enough . in [ s : tree ] we shall show that @xmath52 and @xmath312 can be chosen so that ( [ enough ] ) is satisfied at each step of the construction
. the conditions @xmath325 , as required by lemma [ lem : gaps ] , and @xmath300 , as required by falconer s estimate , will be verified in [ s : tree ] along with the fact that @xmath331 can be chosen large enough to ensure that @xmath310 is close to @xmath7 .
the slits of the next level will be constructed from the previous level using one of two constructions .
the first construction we call the _ liouville construction _ as it uses the liouville convergents of @xmath10 directly to identify new slits .
the second construction , introduced in @xcite , is different .
we call it the _ diophantine construction_. it does not use directly the convergents of @xmath10 , but rather employs a technique to count lattice points in certain strips . in this section ,
we begin with the liouville construction as it is perhaps the main one of the paper . the diophantine construction will be explained in [ s : diophantine ] .
recall that the liouville convergent of a slit @xmath85 indexed by @xmath42 is the vector @xmath348 determined by @xmath349 where @xmath350 note that the height of the liouville convergent satisfies @xmath351 choose @xmath352 so that @xmath353 observe that there are exactly @xmath354 possibilites for @xmath355 .
let @xmath356 consist of children @xmath357 such that @xmath80 forms a basis for @xmath26 together with @xmath89 , i.e. @xmath358 . the next lemma gives a bound on the cross - product of a parent with a child , which recall , is a necessary estimate in the construction of nonergodic directions .
[ lem : max : area ] if @xmath359 for some @xmath360 then @xmath361 where @xmath89 is the liouville convergent of @xmath59 indexed by @xmath42 . from ( [ ieq : lc ] ) we have @xmath362 since @xmath360 , @xmath363 and @xmath364 we have @xmath365 so that @xmath366 .
therefore , @xmath367 the next lemma expresses the key property of slits constructed via the liouville construction .
note that @xmath368 measures how far @xmath369 is from being a reduced fraction ; namely , it is the amount of cancellation between the numerator and denominator . since @xmath370 ( and @xmath371 ) , it is easy to see that @xmath372 .
it is quite surprising that whenever a new slit @xmath106 is constructed via the liouville construction , we have @xmath373 . [
lem : gcd ] for any @xmath374 , we have @xmath373 . hence ,
if @xmath375 , then the inverse slope of @xmath106 has a convergent whose height is either @xmath376 or @xmath377 .
let @xmath378 where @xmath379 now @xmath380 is determined by @xmath381 for some primitive @xmath382 . in terms of the basis given by @xmath89 and @xmath383 we have @xmath384 note that @xmath385 is not divisible by any divisor of @xmath271 , since @xmath235 .
therefore , @xmath386 the second statement follows from lemma [ lem : lc ] .
given @xmath44 we let @xmath387 the next lemma gives a lower bound for the number of children constructed in the liouville construction .
[ lem : liouville ] if @xmath388 then @xmath389 since there are @xmath354 choices for @xmath355 the number of slits in @xmath390 is at least @xmath391 \ge \frac{|w|^r}{|u| } \ge \frac{|w|^{r-1}}{q_k}\ ] ] where @xmath392 was used in the last two inequalities .
now we explain our next general construction , which is accomplished by proposition [ prop : normal ] .
many of the ideas in this section already appeared in @xcite .
again given a parent slit @xmath59 we will construct new slits of the form @xmath357 , where @xmath80 is a loop satisfying certain conditions on its length and cross - product with @xmath59 . not all of these solutions @xmath357 will be used at the next level for it may happen that some of these will not themselves determine enough further slits . in other words
, we will only use some of the slits @xmath357 of the parent @xmath59 and the ones used will be called the _ children _ of @xmath59 .
it will be encumbent to show that there are enough children at each stage in order to obtain lower bounds on the hausdorff dimension of the cantor set of [ s : cantor ] .
assume parameters @xmath393 be given .
in later sections they will each have a dependence on the slit so they are not to be thought of as absolute constants .
we say a slit @xmath59 is @xmath394-_good _ if its inverse slope has a convergent of height @xmath259 satisfying @xmath395 .
let @xmath396 be the collection of slits of the form @xmath357 where @xmath397 satisfies @xmath398 and @xmath399 notice the right hand inequality gives an upper bound for the cross product of @xmath59 with @xmath357 .
the next lemma gives a lower bound for the number of such @xmath357 constructed from good slits @xmath59 .
[ lem : good ] there is a universal constant @xmath400 such that @xmath401 for any @xmath394-good slit @xmath59 and @xmath402 . by [ ch1,thm.3
] , the number of primitive vectors satisfying @xmath403 is at least @xmath404 where @xmath405 is some universal constant .
one needs to assume @xmath406 , but this hypothesis was shown to be redundant in @xcite .
indeed , by [ ch2,thm.4 ] we can take @xmath407 the angle , by which we mean the distance between inverse slopes , between any two solutions @xmath408 to ( [ def : delta_1 ] ) is at least @xmath409 take an interval @xmath410 of length @xmath411 centered at the inverse slope of @xmath59 and divide it into @xmath412 equal subintervals .
the inequality above says that there is at most one solution @xmath80 whose inverse slope lies in each subinterval .
thus , by discarding at most @xmath412 of these solutions , namely those with inverse slopes in @xmath410 , we can ensure that the remaining solutions satisfy @xmath413 these solutions satisfy ( [ def : delta ] ) since @xmath414 let @xmath415 .
we may clearly assume @xmath416 so that @xmath417 . since @xmath402 , there are at least @xmath418 solutions to ( [ def : delta_1 ] ) .
of these , at least one satisfies ( [ def : delta ] ) .
therefore , the number of primitive vectors satisfying ( [ def : delta ] ) is at least @xmath419 [ lem : good : children ] let @xmath59 be an @xmath394-good slit . then every @xmath420 is @xmath421-good , but not @xmath422-good .
let @xmath423 .
note that @xmath80 is a convergent of ( the inverse slope of ) @xmath106 since , writing @xmath378 and @xmath160 , we have @xmath424 and we can use ( [ ieq : cf2 ] ) . let @xmath231 be the height of the next convergent of @xmath106 . then by ( [ ieq : cf1 ] ) @xmath425 so that @xmath426 from the left hand side above , the fact that @xmath427 and @xmath428 , we have @xmath429 now , from the right hand side of ( [ ieq : cfc ] ) , we have @xmath430 this shows that @xmath106 is @xmath421-good . since @xmath259 and @xmath231 are the heights of consecutive convergents of @xmath106 ( and since @xmath431 )
it follows that @xmath106 is not @xmath422-good . in this subsection , we assume @xmath273 is fixed and set @xmath432 the choice of the parameter @xmath187 will depend on considerations in [ s : init ] and will be specified there , by ( [ def : n ] ) .
given @xmath273 , we set @xmath433 it will also be convienent to set @xmath434 a slit @xmath59 is @xmath435-normal if it is @xmath436-good for all @xmath437 $ ] where @xmath438 is determined by @xmath439 .
equivalently , @xmath59 is @xmath435-normal if and only if for all @xmath437 $ ] we have @xmath440\neq\emptyset.\ ] ] where @xmath441 denotes the collection of heights of the convergents of the inverse slope of @xmath59 .
the following gives a sufficient condition for a slit to be normal .
[ lem : n-good=>normal ] let @xmath59 be a slit such that @xmath442 where @xmath443 are consecutive elements of @xmath444 for some @xmath273 . if @xmath59 is @xmath445-good then it is @xmath435-normal .
suppose on the contrary that @xmath59 is @xmath446-good but not @xmath435-normal . implies @xmath438 in the definition of normality .
] let @xmath229 be the convergent of the inverse slope of @xmath59 with maximal height @xmath447 .
since @xmath59 is @xmath445-good , we have @xmath448 let @xmath231 the height of the next convergent .
if @xmath449 then ( [ def : normal ] ) is satisfied by @xmath259 for all @xmath450 $ ] , and by @xmath231 for all @xmath451 $ ] . since @xmath59 is not @xmath435-normal we must have @xmath452 note that @xmath453 writing @xmath85 we have @xmath454 so that @xmath455 from which it follows , by ( [ ieq : cf2 ] ) , that @xmath456 is a convergent of @xmath10 , say @xmath457 since , by ( [ ieq : cf1 ] ) , @xmath458 we have @xmath459 from which it follows that @xmath460 . since @xmath461 , we must have @xmath462 . hence , @xmath463 so that @xmath464 since @xmath465 , we have @xmath466 so that @xmath467 which contradicts the hypothesis on @xmath468 . given a slit @xmath59 let @xmath469 our goal , proposition [ prop : normal ] , is to develop hypotheses on an @xmath435-normal slit @xmath59 that ensures that among the slits @xmath470 lots of them are @xmath471-normal .
more specifically we wish to show that under suitable hypotheses , an @xmath435-normal slit @xmath59 determines lots of @xmath471-normal @xmath472 where @xmath78 and @xmath473 if @xmath106 is @xmath471-normal and satisfies ( [ def : child ] ) then it will be called a _ child _ of @xmath59 .
the main task will be to bound the number of @xmath106 that satisfy ( [ def : child ] ) but are not @xmath471-normal .
we begin with a pair of lemmas that are essentially a consequence of normality .
[ lem : t_1 ] suppose @xmath59 is @xmath435-normal .
let @xmath89 be the convergent of the inverse slope with maximum height @xmath474 and @xmath259 the height of the next convergent .
define @xmath475 by @xmath476 and @xmath477 by @xmath478 then @xmath479 and @xmath480 .
the first inequality is a consequence of the case @xmath481 in the definition of normality the left hand part of the second inequality follows from the defintion of @xmath259 , while the right hand follows from @xmath435-normality because there would otherwise be a @xmath482 for which ( [ def : normal ] ) fails . [ lem : t_1<n ] suppose @xmath420 is not @xmath471-normal and again letting @xmath483 be the convergent of the inverse slope with maximum height @xmath484 and @xmath231 the next convergent define @xmath485 by @xmath486 and @xmath487 by @xmath488 .
. then @xmath490 .
we first note again that @xmath491 by definition .
now @xmath492 since if @xmath493 then ( [ def : normal ] ) is satisfied by @xmath494 for all @xmath495 $ ] , and by @xmath231 for all @xmath496 $ ] , contrary to the assumption that @xmath106 is not @xmath471 normal .
if @xmath497 then lemma [ lem : n-good=>normal ] applied to the slit @xmath106 , with @xmath498 in place of @xmath435 , implies that @xmath106 is @xmath471-normal , contrary to assumption .
[ lem : strips ] suppose @xmath499 satisfies the conditions of lemma [ lem : t_1<n ] .
let @xmath500 .
let @xmath483 be the convergent of @xmath106 as above then @xmath483 determines a ( nonzero ) integer @xmath501 such that @xmath502 moreover , @xmath503 .
write @xmath472 and recall that since @xmath504 ( as in the proof of lemma [ lem : good ] ) @xmath80 is a convergent of @xmath106 .
let @xmath158 be the next convergent of @xmath106 after @xmath80 .
since @xmath505 we either have @xmath506 or @xmath483 comes after @xmath158 in the continued fraction expansion of @xmath106 . in any case , we have @xmath507 for some _ nonnegative _ integers @xmath508 with @xmath509 . since @xmath510 we have @xmath511 on the other hand ,
@xmath512 this proves the first part . by the first inequality in ( [ ieq : cfc ] ) ,
@xmath513 so that @xmath514 by lemma [ lem : t_1<n ] and since @xmath515 .
this proves the second part .
suppose @xmath516 also satisfies ( [ def : child ] ) and is also not @xmath498-normal and satisfies @xmath517 .
let @xmath518 be the convergent of @xmath519 with maximal height @xmath520 .
suppose further that it determines the same integer @xmath203 determined by @xmath483 as in lemma [ lem : strips ] .
then we say @xmath483 and @xmath518 belong to the same _ strip_. the number of strips is bounded by the number of possible values for @xmath203 .
thus , by lemma [ lem : strips ] , the number of strips is bounded by @xmath521 now suppose @xmath522 belong to the same strip .
we say @xmath483 and @xmath518 lie in the same _ cluster _ if they differ by a multiple of @xmath89 .
[ lem : same : cluster ] if @xmath523 then they belong to the same cluster .
since @xmath522 determine the same @xmath203 , lemma [ lem : strips ] and the fact that @xmath524 implies @xmath525 so that writing @xmath526 where @xmath527 we have @xmath528 which implies @xmath529 is a convergent of @xmath59 . since @xmath530 , we have @xmath531 , by definition of @xmath89 .
now suppose @xmath532 .
we will arrive at a contradiction . since @xmath89 is a convergent of @xmath59 coming after @xmath529 , @xmath533 which together with ( [ ieq : width : strip ] ) implies @xmath534 so that @xmath535 contradicting the definition of @xmath89 .
we conclude that @xmath536 , so that @xmath522 differ by a multiple of @xmath89 .
that is , they belong to the same cluster .
pick a representative from each cluster . to bound the number of clusters we bound the number of representatives .
since @xmath537 and the difference in height of any two representatives is greater than @xmath538 , the number of clusters is bounded by ( since @xmath465 ) @xmath539 to bound for the number of @xmath483 in each cluster we need an additional assumption .
[ lem : cluster : size ] suppose @xmath540 ( independent of @xmath483 within the cluster )
. then the number of elements in the cluster is bounded by @xmath541 lemma [ lem : strips ] implies for any @xmath522 in the cluster @xmath542 where @xmath543 is the smallest possible within the cluster . on the other hand , @xmath544 so that @xmath545 by definition @xmath546 is a multiple of @xmath89 .
to get the desired bound , using the assumptions @xmath547 and @xmath548 , it remains to show that @xmath549 to see this note that if @xmath493 then since @xmath540 @xmath550 whereas if @xmath551 then @xmath552 we shall now apply our lemmas to show that , under suitable hypotheses on an @xmath435-normal slit @xmath59 there are lots of children , i.e. @xmath471-normal slits @xmath106 satisfying ( [ def : child ] ) . [
prop : normal]suppose @xmath59 is an @xmath435-normal slit satisfying @xmath553 where @xmath443 are consecutive elements of @xmath444 .
suppose further that @xmath554 then the number of @xmath106 satisfying ( [ def : child ] ) that are @xmath471-normal is at least @xmath555 let @xmath475 be the parameter associated to the convergent @xmath89 of @xmath59 as in ( [ lem : t_1 ] ) .
there are two cases . if @xmath556 then @xmath59 is @xmath557-good , so that lemma [ lem : good ] implies @xmath59 has at least @xmath558 @xmath472 satisfying ( [ def : child ] ) .
moreover , by lemma [ lem : good : children ] each @xmath106 constructed is @xmath559-good . since @xmath560 by the choice of @xmath561 , every such @xmath106 is @xmath562-good . moreover
, since each @xmath106 has length at most @xmath563 , lemma [ lem : n-good=>normal ] implies each @xmath106 constructed is @xmath471-normal . note that the number in ( [ double ] ) is twice as many as we need
now consider the case @xmath564 . in this case
@xmath59 is @xmath565-good , so that lemma [ lem : good ] implies @xmath59 has at least @xmath566 @xmath106 satisfying ( [ def : child ] ) .
moreover , lemma [ lem : good : children ] implies each child @xmath106 constructed is @xmath567-good , and since @xmath568 again , by the choice of @xmath561 , this means @xmath106 is @xmath569-good .
moreover , the parameter @xmath485 associated to the convergent @xmath483 of each such @xmath106 satisfies @xmath540 . applying lemmas [ lem : strips ] , [ lem : same : cluster ] and [ lem : cluster : size ] we conclude the number of @xmath106 constructed that are not @xmath471-normal is at most the product of the bounds given in ( [ bound : strips ] ) , ( [ bound : clusters ] ) , and lemma ( [ lem : cluster : size ] ) , i.e. @xmath570 which is at most half the amount in ( [ double ] ) since ( [ ieq : normal ] ) holds .
in this section we specify some parameters that need to be fixed before the construction of the tree of slits can begin .
in particular , we shall specify the initial slit .
we shall also specify the type of construction that will be used at each level to find the slits of the next level . given @xmath210
we first choose @xmath547 so that @xmath571 then choose @xmath572 so that @xmath573 it will be convenient to set @xmath574 and let @xmath575 we set @xmath576 and let @xmath577 be given by ( [ def : n ] ) .
we assume that @xmath444 , which was defined in ( [ def : ell_n ] ) , has infinitely many elements , for if @xmath444 were finite , then @xmath10 is diophantine and this case has already been dealt with in @xcite
. our argument would simplify considerably if we assume @xmath444 is finite and it would essentially reduce to the one given in @xcite . now choose @xmath578 large enough so that @xmath579 [ lem : init : slit ] there is a slit @xmath580 such that @xmath581 and @xmath582 let @xmath583 be any slit such that @xmath584 . choose @xmath585 with minimal height satisfying the first inequality in ( [ ieq : init : slit ] ) .
lemma [ lem : gcd ] implies @xmath581 . let @xmath89 be the liouville convergent of @xmath59 indexed by @xmath208 . its height @xmath586 .
since consecutive elements in @xmath587 differ by @xmath588 , we have @xmath589 since @xmath590 .
choose @xmath293 satisfying the conditions of lemma [ lem : init : slit ] and let it be fixed for the rest of this paper .
it is the unique slit of level @xmath6 .
note that the choice of @xmath208 in ( [ def : k_0 ] ) gives various lower bounds on the length of @xmath293 , by virtue of the first inequality in ( [ ieq : init : slit ] ) .
for example , since @xmath591 , the first relation in ( [ def : k_0 ] ) implies @xmath592 next , we shall specify for each level @xmath593 the type of construction that will be applied to the slits of level @xmath54 to construct slits of the next level .
( the same type of construction will be applied to all slits within the same level . )
we shall define indices @xmath594 for each @xmath595 with @xmath596 and for @xmath597 such that whenever @xmath598 are consecutive elements of @xmath444 we have ( see lemma [ lem : indices](i ) below ) @xmath599 for @xmath600 we use the construction described in [ s : liouville ] , while for all other @xmath54 we use the techniques described in
[ s : diophantine ] . the precise manner in which these types of constructions will be applied
is described in the next subsection .
the primary role of these indices is to ensure that various conditions on the lengths of all slits in some particular level are satisfied .
( see lemma [ lem : range ] . ) specifically , the conditions in lemmas [ lem : gcd ] and [ lem : liouville ] are needed for the levels @xmath601 and those in proposition [ prop : normal ] are needed for the levels @xmath602 .
it will also be important that the number of levels between @xmath603 and @xmath604 be bounded ( lemma [ lem : indices].ii ) whereas the number between @xmath604 and @xmath605 ( or between @xmath605 and @xmath606 ) will generally not be bounded .
let @xmath607 and for @xmath608 set @xmath609\ ] ] so that the lengths of all slits of level @xmath54 lie in @xmath610 , by ( [ ieq:|w_j+1| ] ) .
[ lem : h_j ] for all @xmath593 @xmath611 the condition @xmath612 is equivalent to @xmath613 which is implied by @xmath614 which in turn is implied by ( [ ieq : w_0>5 ] ) .
the choice of the indices @xmath594 will depend on the position of @xmath610 relative to that of the following intervals : @xmath615 here , again , @xmath616 is the element in @xmath444 immediately after @xmath42 .
these intervals overlap nontrivially and the overlap can not be too small in the sense that there are at least three consecutive @xmath610 s contained in it .
[ lem : ikcd3 ] for any @xmath595 with @xmath596 @xmath617 note that @xmath618 sends @xmath619 to a nonnegative integer and @xmath620 for any @xmath259 the image of @xmath621 under @xmath622 contains exactly @xmath623 integers , all of them nonnegative if @xmath624 ; or equivalently , if @xmath625 . under this condition
, the fact in lemma [ lem : h_j ] that @xmath626 implies @xmath627 since @xmath628 and @xmath629 , we have @xmath630 and since @xmath631 , ( [ ieq : ikcd3 ] ) follows . by virtue of the fact that the quantity in ( [ ieq : ikcd3 ] ) is at least one
, we can now give two equivalent definitions of the index @xmath594 .
for @xmath598 consecutive elements of @xmath444 with @xmath596 , let @xmath632 note that @xmath633 and that @xmath634 is not defined .
the main facts about these indices are expressed in the next two lemmas .
[ lem : indices ] for any @xmath635 , @xmath596 1 .
@xmath636 2 .
@xmath637 for ( i ) we note that @xmath638 so the first inequality follows by the ( second ) definition of @xmath639 . from the first definitions of @xmath604 and @xmath605 , we see that the second inequality is a consequence of lemma [ lem : ikcd3 ] .
the third inequality follows by comparing the second definitions of @xmath605 and @xmath606 and noting that @xmath640 . for ( ii ) first note that @xmath641 by lemma [ lem : h_j ] and the second definition of @xmath603 .
thus , we have @xmath642 where @xmath643 .
the second definition of @xmath604 now implies @xmath644 .
[ lem : range ] for any slit @xmath59 of level @xmath54 we have 1 .
@xmath645 2 .
@xmath646 by definition , @xmath647 and @xmath648 , giving the first implication in ( i ) . since @xmath649
we have @xmath650 so that @xmath651 .
this , together with @xmath652 , implies the second implication in ( i ) . for ( ii ) note that ( [ ieq : ikcd3 ] ) implies @xmath653 , giving the first implication , while the second implication follows from @xmath654
in this section we specify exactly how the slits of level @xmath294 are constructed from the slits of level @xmath54 . as before
, we refer to any slit constructed from a previously constructed slit @xmath59 as a _ child _ of @xmath59 .
the parameters @xmath52 and @xmath312 are also specified in this section . at each step
, we shall verify that the choice of @xmath52 and @xmath312 is such that all cross - products of slits of level @xmath54 with their children are @xmath655 while the number of children is at least @xmath656 , as required by ( [ enough ] ) in [ s : cantor ] .
depending on the type of construction to be applied , there will be various kinds of hypotheses on all slits within a given level that we need to verify .
these hypotheses can be one of two kinds .
the first kind involve inequalities on lengths of slits and these will always be satisfied using lemma [ lem : range ] .
we will not check these hypotheses explicitly .
the second kind is more subtle and involve conditions related to the continued fraction expansions of the inverse slopes of slit directions .
the fact that we need such hypotheses on slits is evident from lemma [ lem : good ] , which is one of the main tools we have for determining whether a slit will have lots of children .
one of the main tasks of this section will be to check the required hypotheses of the second kind at each step .
for the levels between consecutive indices of the form @xmath594 , these hypotheses will hold by virtue of the results in [ s : liouville ] and [ s : diophantine ] .
special attention is needed to check the relevant hypotheses of the second kind for the levels @xmath657 when the type of construction used to find the slits of the next level changes .
in what follows , it will be implicitly understood that @xmath598 denote consecutive elements of @xmath444 , with @xmath596 .
if @xmath658 , then @xmath659 will denote the element of @xmath444 immediately before @xmath42 .
for the levels @xmath54 satisfying @xmath660 , the slits of level @xmath294 will be constructed by applying lemma [ lem : liouville ] to all slits of level @xmath54 .
in other words , the slits of level @xmath294 consist of all slits @xmath661 where @xmath59 is a slit of level @xmath54 and @xmath80 is a loop such that @xmath472 .
recall that an initial slit @xmath293 has been fixed using lemma [ lem : init : slit ] .
lemma [ lem : max : area ] implies the cross - products of @xmath293 with its children are all less than @xmath662 , while lemma [ lem : liouville ] implies the number children is at least @xmath663 .
therefore , we set @xmath664 for the levels @xmath665 , we set @xmath666 [ lem : lr ] for @xmath667 , every slit @xmath59 of level @xmath54 satisfies @xmath668 .
moreover , if @xmath669 then the cross - products of each slit of level @xmath54 with its children are less than @xmath52 and the number of children is at least @xmath656 .
since all slits of level @xmath54 were obtained via the liouville construction , the first part follows from the first assertion of lemma [ lem : gcd ] .
suppose @xmath59 is a slit of level @xmath54 with @xmath665 .
lemma [ lem : max : area ] now implies the cross - products of @xmath59 with its children are less than @xmath670 , and the number of children is at least @xmath671 , by lemma [ lem : liouville ]
. it will be convenient to set @xmath672 [ lem : lr->dr ] every slit of level @xmath605 is @xmath673-normal .
let @xmath59 be a slit of level @xmath605 .
since @xmath674 , we have @xmath675 by lemma [ lem : lr ] , we have @xmath668 and since @xmath59 was obtained via the liouville construction , lemma [ lem : gcd ] implies the inverse slope of @xmath59 has a convergent with height between @xmath676 and @xmath677 , or , by the above , between @xmath678 and @xmath538 .
this means @xmath59 is @xmath679-good , and therefore , @xmath673-normal , by lemma [ lem : n-good=>normal ] . for the levels
@xmath54 satisfying @xmath680 , the slits of level @xmath294 will be constructed by applying proposition [ prop : normal ] with the parameter @xmath681 to all slits @xmath59 of level @xmath54 . in other words ,
the slits of level @xmath294 consist of all @xmath471-normal children of all slits of level @xmath54 , where @xmath682 . for the levels @xmath680
, we set @xmath683 [ lem : dr ] for @xmath602 , every slit @xmath59 of level @xmath54 is @xmath684-normal .
morevover , if @xmath685 then the cross - products of each slit of level @xmath54 with its children are less than @xmath52 and the number of children is at least @xmath656 .
the case @xmath686 of the first assertion follows from lemma [ lem : lr->dr ] while the remaining cases follow from proposition [ prop : normal ] . for children constructed via proposition [ prop : normal ] applied to an @xmath435-normal slit ,
the cross - products are less than @xmath687 , which is @xmath52 if @xmath681 .
the number of children is at least @xmath688 provided we verify that the inequality ( [ ieq : normal ] ) holds , i.e. if @xmath689 to check this inequality , we first note that @xmath690 so that @xmath691 since @xmath692 , by the first relation in ( [ def : m ] )
. next , we note that it is enough to check ( [ ieq : normal:2 ] ) in the case @xmath686 since the left hand side increases by a factor @xmath693 as @xmath54 increments by one , while the right hand side increases by a factor @xmath694 .
moreover , since @xmath695 , ( [ ieq : normal:2 ] ) in the case @xmath686 follows from @xmath696 , which is guaranteed by the second term in ( [ def : k_0 ] ) .
[ lem : dr->br ] every slit @xmath59 of level @xmath606 is @xmath697-good .
let @xmath59 be a slit of level @xmath606 .
lemma [ lem : dr ] implies that @xmath59 is @xmath435-normal for some @xmath698 . by the case @xmath481 in the definition of normality , this means @xmath59 is @xmath699-good , i.e. its inverse slope has a convergent whose height is between @xmath700 and @xmath538 . since @xmath698
the height of this convergent is between @xmath701 and @xmath538 .
hence , @xmath59 is @xmath697-good . for the levels
@xmath54 satisfying @xmath702 , the slits of level @xmath294 will be constructed by applying lemma [ lem : good ] to all slits @xmath59 of level @xmath54 with the parameters @xmath703 in other words , the slits of level @xmath294 consist of all slits of the form @xmath357 where @xmath59 is a slit of level @xmath54 and @xmath704 where @xmath435 and @xmath705 are the parameters given in ( [ alpha : beta ] ) . for the levels @xmath702
, we set @xmath706 [ lem : br ] for @xmath707 , every slit @xmath59 of level @xmath54 is @xmath708-good .
morevover , if @xmath709 then the cross - products of each slit of level @xmath54 with its children are less than @xmath52 and the number of children is at least @xmath710 .
first we note that every slit @xmath59 of level @xmath54 is @xmath394-good , where @xmath435 and @xmath705 are the parameters given in ( [ alpha : beta ] ) .
indeed , for @xmath711 this follows from lemma [ lem : dr->br ] while for @xmath712 it follows from lemma [ lem : good : children ] .
lemma [ lem : indices].ii and the third relation in ( [ def : k_0 ] ) imply @xmath713 from which we see that the first assertion holds . for children constructed via lemma [ lem : good ] applied to an @xmath394-good slit , the cross - products are less than @xmath687 , which is @xmath655 , since @xmath714 . and since @xmath715 , the number of children is at least @xmath716 giving the second assertion .
finally , for the levels @xmath717 with @xmath658 , we set @xmath718 [ lem : br->lr ] for any slit @xmath59 of level @xmath717 with @xmath658 , the cross - products of @xmath59 with its children are less than @xmath52 and the number of children is at least @xmath656 .
suppose @xmath59 is a slit of level @xmath604 with @xmath658 .
the case @xmath717 of lemma [ lem : br ] implies @xmath59 is @xmath708-good .
let @xmath89 be the liouville convergent of @xmath59 indexed by @xmath42 .
by lemma [ lem : lc ] the height @xmath231 of the next convergent is @xmath719 since @xmath59 is @xmath708-good , we must have @xmath720 so that , by lemma [ lem : max : area ] the cross - products of @xmath59 with its children are @xmath721 by lemma [ lem : liouville ] , the number of children is at least @xmath722 .
the construction of the tree of slits is now complete .
we gather the definitions of @xmath52 and @xmath312 ( for @xmath608 ) in the table below .
@xmath723 & \delta_j & & \frac{4\rho^{n'}}{q_{\tilde{k } } } & & \frac{8\rho^{n'}}{q_{\tilde{k } } } & & \frac{4}{q_k } & & \frac{2\rho^{n'}}{q_kr^{j - j^d_k } } \\[12pt ] & \rho_j & & \frac{c_0}{2 } & & \frac{q_{\tilde{k}}}{8\rho^{n'}q_k } & & \frac{1}{4 } & & \frac{c_0}{2\rho^{n'+1 } } \\[12pt ] & \rho_j\delta_j & & \frac{2c_0\rho^{n'}}{q_{\tilde{k } } } & & \frac{1}{q_k } & & \frac{1}{q_k } & & \frac{c_0/\rho}{q_kr^{j - j^d_k } } \\\end{aligned}\ ] ] from the fourth relation in ( [ def : k_0 ] ) we see that @xmath726 . for @xmath660
we have @xmath727 and since @xmath728 , by the definition of @xmath604 , we have @xmath729 , from which it easily get @xmath730 and , in particular , @xmath300 . for @xmath731 the expression @xmath732 increases faster than @xmath733 decreases , so it is enough to check the case @xmath686 , for which , by the above , we have @xmath734 .
next , we obtain the lower bound on the hausdorff dimension of @xmath314 .
recall the expression for the local hausdorff dimensions @xmath310 given in ( [ formula : d_j ] ) .
the next lemma shows it is close to @xmath7 by the choices made in
[ s : init ] . by ( [ eq : local ] )
it is enough to show that the term ( [ eq : num ] ) and both of the terms in ( [ eq : den ] ) are bounded by @xmath736 . by the choice of @xmath737 in ( [ def : m ] )
, it would be enough to show that each term is bounded by @xmath738 for all large enough @xmath54 .
it will be convenient to write @xmath739 as an abbreviation for @xmath740 .
we consider the expression ( [ eq : den ] ) first . using ( [ ieq : w_0>5 ] ) and the fact that @xmath741 we see that the first term in ( [ eq : den ] ) satisfies @xmath742 from the last row of the table , we see that for @xmath743 we have @xmath744 while for @xmath745 we have @xmath746 then , in the second case , we have @xmath747 since @xmath748 ; in the first case the left hand side above is @xmath749 .
we now turn to the expression ( [ eq : num ] ) . for @xmath660
we have @xmath750 next consider @xmath680 .
using @xmath751 , we have @xmath752 finally , we turn to the possibility that @xmath753 ( @xmath754 ) . since @xmath755 , we have @xmath756 and the lemma follows .
it suffices to check that @xmath758 for in that case , every sequence @xmath759 constructed above satisfies ( [ ieq : sumx ] ) and @xmath757 , by theorem [ thm : sumx ] .
we break the sum into three intervals : @xmath707 , @xmath760 , and @xmath761 .
the construction of the set @xmath314 as well as the lower bound @xmath43 estimate on its hausdorff dimension remains valid for any irrational @xmath10 .
( note that when @xmath769 , @xmath314 can not be a subset of @xmath770 since the latter has hausdorff dimension @xmath6 ) . on the other hand
the fact that @xmath771 implies @xmath772 , by [ ch2,prop .
therefore , @xmath773 for all irrational @xmath10 .
the opposite inequality follows from a more general result in @xcite .
lastly , when @xmath774 , the set @xmath65 is countable , so that its hausdorff dimension vanishes . mmm y. cheung , _ hausdorff dimension of the set of nonergodic directions_. with an appendix by m. boshernitzan . ann . of math .
( 2 ) * 158 * ( 2003 ) , no .
2 , 661678 .
y. cheung , _ slowly divergent geodesics in moduli space , _ conform .
* 8 * ( 2004 ) , 167189 .
y. cheung , _ hausdorff dimension of the set of singular pairs_. ann . of math .
, to appear .
` arxiv:07094534 ` y. cheung , a. eskin , _ slow divergence and unique ergodicity _ ,
preprint . ` arxiv:0711.0240v1 ` a.eskin , h.masur , _ asymptotic formulas on flat surfaces _ , erg .
* 21 * 443478 .
k. falconer , _ fractal geometry .
mathematical foundations and applications , _ john wiley & sons ltd . , chichester , 1990 .
khintchin , _ continued fractions _ , university of chicago press , 1964 .
english translation .
first russian edition published 1935 .
h. keynes , d. newton , _ a minimal non uniquely ergodic interval exchange _
, math z * 148 * ( 1976 ) , 101106 .
h.masur _ the growth rate of trajectories of a quadratic differential _ erg .
10 * ( 1990 ) 151 - 176 h. masur , _ hausdorff dimension of the set of nonergodic foliations of a quadratic differential_. duke math . j.
* 66 * ( 1992 ) , no .
3 , 387442 . j.w .
milnor , _ dynamics in one complex variable _ , vieweg , 1999 , 2000 ; princeton u. press , 2006 .
h. masur , j. smillie , _ hausdorff dimension of sets of nonergodic foliations _
, ann . of math .
* 134 * ( 1991 ) , 455-543 .
h. masur , s. tabachnikov , _ rational billiards and flat structures_. handbook of dynamical systems , vol .
1a , 10151089 , north - holland , amsterdam , 2002 .
r. prez marco , _ sur les dynamiques holomorphes non linarisables et une conjecture de v. i. arnold .
english summary ) [ nonlinearizable holomorphic dynamics and a conjecture of v. i. arnold ] _ ann .
sci . ecole norm .
( 4 ) * 26 * ( 1993 ) , no . 5 , 565644 . h. poincar , _ sur un mode nouveau de reprsentation gomtrique des formes quadratiques dnies et indnies , _ ( 1880 ) in oeuvres compltes de poincar , tome v , ( 1952 ) , 117183 .
w. veech , _ strict ergodicity in zero dimensional dynamical systems and the kronecker - weyl theorem @xmath775_. trans .
* 140 * ( 1969 ) , 133 .
w. veech , _ teichmller curves in moduli space , eisenstein series and an application to triangular billiards_. invent .
* 97 * ( 1989 ) , no .
3 , 553583 .
y. vorobets _ periodic geodesics on translation surfaces _ contemporary math . * 385 * , amer . math .
soc . , providence , ri , ( 2005 ) . | given an irrational @xmath0 , we consider billiards in the table @xmath1 formed by a @xmath2 rectangle with a horizontal barrier of length @xmath3 with one end touching at the midpoint of a vertical side .
let @xmath4 be the set of @xmath5 such that the flow on @xmath1 in direction @xmath5 is not ergodic .
we show that the hausdorff dimension of @xmath4 can only take on the values @xmath6 and @xmath7 , depending on the summability of the series @xmath8 where @xmath9 is the sequence of denominators of the continued fraction expansion of @xmath10 . more specifically , we prove that the hausdorff dimension is @xmath7 if this series converges , and @xmath6 otherwise .
this extends earlier results of boshernitzan and cheung . |
collectively , pulsars emit over essentially the entire electromagnetic spectrum .
however , most pulsars are observed only in a relatively narrow range , @xmath0@xmath1 .
the basic properties determined for most pulsars are the period , @xmath2 , and the period derivative , @xmath3 , which determine the magnetic field , @xmath4 , the characteristic age @xmath5 and the spin - down power @xmath6 . also available for most pulsars
are the integrated pulse profile , the position angle of the linear polarization , which determines the inclination angle between the magnetic and rotational axes , and the dispersion measure , which provides an estimate of the distance to the pulsar . based on the location in the @xmath2@xmath3 plane
, pulsars are classified as young ( @xmath7 ) , old , recycled ( or millisecond ) , or magnetar .
it is remarkable that despite the large ranges in @xmath2 , @xmath8 , the variations in pulse profile between classes are similar to those within classes .
this suggests that the radio emission mechanism can not be sensitive to either @xmath2 or @xmath8 .
the source of the radio emission appears to be between inner and outer gaps , seemingly far from these sites where parallel electric fields are thought to accelerate primary particles and trigger pair cascades .
only a small fraction of radio pulsars are also observed at high energies , in the x - ray and gamma - ray ranges , which emissions should provide a direct signature of particle acceleration and pair creation .
however , there remains a major uncertainty in the location of the source region , with both inner - gap and outer - gap models considered viable .
high - energy emission mechanism are summarized in section 2 and radio emission mechanisms are summarized in section 3 .
the identification of source regions is discussed in section 4 .
the high - energy emission mechanisms that need to be considered for pulsars are synchrotron emission , curvature emission and inverse compton emission ( ic ) , all of which are relatively well known . a less familiar , resonant form of ic emission ( ric ) is important .
a preliminary point is that in a relativistic quantum treatment , the energy of an electron has discrete values @xmath9 where @xmath10 is the component of the momentum along the field lines , and @xmath11 is the landau quantum number . in a pulsar
the lifetime of the excited states is very short , and all electrons quickly relax to their ground state , @xmath12 . synchrotron emission occurs only if highly relativistic pairs are created with @xmath13 , corresponding to @xmath14 , @xmath15 , @xmath16 pitch angle . the emission is strongly concentrated around an angle @xmath17 , with a broad frequency spectrum peaked at @xmath18 , where @xmath19 is the cyclotron frequency .
the power radiated is @xmath20 , and the radiation is highly linearly polarized , @xmath21 , in the direction perpendicular to the projection of the magnetic field on the plane of the sky . for a power - law energy spectrum of radiating particles , @xmath22 ,
the intensity spectrum is a power law , @xmath23 .
curvature emission by an individual electron with lorentz factor @xmath24 moving along ( @xmath12 ) a magnetic field line with radius of curvature @xmath25 is strongly concentrated around @xmath26 , with a broad frequency spectrum peaked at @xmath27 , @xmath28 .
the power radiated is @xmath29 , and the radiation is highly linearly polarized in the direction perpendicular to the plane containing the curved magnetic field line . for a power - law energy spectrum of radiating particles the intensity spectrum is a power law with index @xmath30 .
ic emission is actually thomson scattering by highly relativistic electrons .
an initial photon with frequency @xmath31 is scattered into a final photon with frequency @xmath32 propagating nearly along the direction of the initial electron .
the power radiated is @xmath33 , where @xmath34 is the thomson cross section and @xmath35 is the energy density in the initial photons .
the polarization has no characteristic value in general , and the spectrum is synchrotron - like .
relativistic quantum effects sharply reduce the cross section , from the thomson to the klein - nishina value , for @xmath36 .
ric emission is thomson scattering by highly relativistic electrons in a magnetic field such that the wave frequency , @xmath31 , is near the relativistic cyclotron frequency , @xmath37 . when this condition is satisfied , the thomson cross section is greatly enhanced , by a factor @xmath38 ( canuto , lodenquai & ruderman 1971 ) dependent on the polarization of the photon ( melrose & sy 1972 ) .
the maximum enhancement factor is @xmath39 , where @xmath40 is the width of the cyclotron resonance .
ric involves scattering in which the initial and final state of the electron is @xmath12 and the ( virtual ) intermediate state is @xmath41 .
the relevant @xmath40 is the inverse lifetime of the first excited state ( herold 1979 ) , which is @xmath42 .
the importance of ric in high - energy emission from pulsars was emphasized by daugherty & harding ( 1989 ) and dermer ( 1990 ) .
there are both inner - gap and outer - gap models for the high - energy emission , and there is no clear consensus on which type of model is to be preferred .
high - energy emission should provide a signature of acceleration and pair production in inner - gap models , provided a small fraction of high - energy photons escape .
recent inner - gap models ( zhang & harding 2000 , hibschman & arons 2001 , arendt & eilek 2002 ) confirm the general features of earlier models , generally favoring a higher fraction of lower energy , @xmath43 , pairs and pair creation extending to greater heights than earlier models
. the efficiency of conversion of the energy in primary particles can be very high , when acceleration is radiation - reaction limited due to curvature emission losses .
ric emission off thermal photons from the surface of the neutron star can be particularly effective ( zhang & harding 2000 ) .
synchrotron emission from the secondary pairs can contribute to the observed high - energy emission , provided the magnetic field is not too strong .
existing models are not satisfactory for relatively strong fields for reasons discussed by zhang & harding ( 2000 ) .
another effect that has not been included is that for @xmath44@xmath45 ( @xmath46 ) , the secondary pairs are created predominantly in their ground state ( weise & melrose 2002 ) or as bound pairs ( usov & melrose 1996 ) , in which cases synchrotron emission is absent . outer - gap models ( e.g , cheng , ho & ruderman 1986 ) involve particle acceleration , high - energy emission and pair production in an outer gap , relatively near the light cylinder .
some of the relevant parameters in the outer gap are quite different from the inner gap : @xmath8 is much weaker , @xmath25 is much larger , and the density of thermal photons from the surface of the star is much lower .
emission of photons with a given energy due to synchrotron emission or curvature emission requires particles with a much higher @xmath47 in the outer gap compared with the inner gap .
the low density of thermal photons implies that they are unimportant as targets for ic emission .
romani ( 1996 ) compared the contributions to the high - energy emission due to the various processes for a particular choice of parameters , as illustrated in figure 1 .
qualitative arguments in favor of outer - gap models are that they lead naturally to the double - peaked profiles of the form observed in some sources , and imply harder spectra from older , more efficient pulsars ( romani 1996 ) .
the sensitivity of the models to various parameters , notably @xmath8 and the inclination angle between rotation and magnetic axes , provides scope for accounting for most of the observed features of high - energy emission from individual pulsars .
the properties of inner - gap and outer - gap models ( cheng & zhang 1999 ) were compared by zhang & harding ( 2000 ) , concentrating on x - ray emission . qualitatively , nonthermal x - ray emission in inner - gap models is attributed to ric emission due to upward - propagating pairs , and in outer - gap models it is attributed to synchrotron emission by downward propagating pairs .
thermal x - ray emission from polar - cap heating is determined by the energy flux of secondary pairs incident on the polar cap , heating it , and this flux is different in the different models .
however , these and other differences have yet to provide definitive arguments that distinguish between the two types of model .
the uncertainty as to whether the source region is in the inner - gap or the outer - gap , or even in the wind ( kirk , skraasen & gallant 2002 ) , is likely to be resolved only when more detailed observational data on the high - energy emission becomes available .
a hint as to how a partial resolution might arise is through an obscuring of the distinction between inner and outer gaps .
inefficient screening of the accelerating electric field ( e.g. , hibschman & arons 2001 ) allows pair production and high - energy emission from an inner gap to extend to greater heights than previously thought , and the dependence of the location of the outer gap on the global current ( hirotani & shibata 2001 ) allows pair production and high - energy emission from an outer gap to occur at lower heights than previously thought . it may be that acceleration and pair production occur at a significant level over a wide range of heights
amongst the many pulsar radio emission mechanisms that have been proposed , only those that involve some form of plasma instability or maser action are mentioned here . there are at least five distinct instability or maser mechanisms currently under consideration .
the two most favorable of these are discussed here : a form of maser curvature emission and relativistic plasma emission ( beam - driven cerenkov instability ) .
the other three are described more briefly .
_ maser curvature emission _ : although initially thought impossible ( blandford 1975 ; melrose 1978 ) , maser curvature emission is possible when either curvature - drift ( luo & melrose 1992 ) or field line torsion ( luo & melrose 1995 ) are taken into the account .
there are two unfavorable features of curvature - drift - driven maser .
first , it is sensitive to @xmath8 , and so can not plausibly be applied to all classes of pulsars .
second , maser emission is possible only for lorentz factors above a relatively high threshold , @xmath48 .
however , like all relevant instability mechanisms , growth requires that the particle distribution function be an increasing functions of lorentz factor , @xmath49 , and this condition is satisfied only for @xmath43 in current models for pair cascades ( hibschman & arons 2001 ; arendt & eilek 2002 ) .
the torsion - driven maser requires that the field line not be confined to a plane , so that it has two different radii of curvature in orthogonal directions .
there are two possible sources of torsion .
one is rotation , which causes a sweeping back of the field lines .
this is significant only near the light cylinder .
the other is a non - dipolar component of the field , which tends to produce torsion in the inner magnetosphere .
for example , a quadrupolar component would provide torsion in general , with the ratio of the radii of curvature increasing linearly with radial distance from the center of the star . like the curvature - drift - driven maser , growth is possible only above a minimum @xmath47 , but in this case the threshold , @xmath50 ( luo & melrose 1995 ) , is compatible with @xmath49 from pair cascade models .
a strong argument in favor of torsion - driven maser curvature emission is that the mechanism depends sensitively only on the primary radius of curvature , and is not particularly sensitive to the ratio of the radii of curvature .
hence the properties of the emission is determined by the dipolar component of the magnetic field , which is plausibly similar in all pulsars . as a consequence ,
the mechanism can account naturally for the remarkable similarity in the frequency range for radio emission from all classes of pulsars .
another argument in favor of the mechanism is that growth occurs only in one linear polarization ( that orthogonal to the dominant polarization in incoherent curvature emission ) , and this is consistent with observational evidence on the plane of polarization ( radhakrishnan & deshpande 2001 ) .
_ relativistic plasma emission _ : some form of beam - driven wave growth has long been favored as the pulsar radio emission mechanism , but various difficulties with it have led to ongoing doubts about its viability .
generically , a beam - driven cerenkov instability requires ( a ) that the growing waves have refractive index greater than unity ( phase speed @xmath51 ) , and ( b ) that the distribution of particles includes a counterstreaming motion such that one component is streaming through another at greater than the phase speed of the growing wave .
effective growth requires that the growth rate for the instability be greater than effective loss rates for the growing waves .
seemingly plausible beams are primary particles streaming through secondary pairs , or electrons streaming relative to positrons to provide the required current density , but for both the growth rate is far too small .
one seems to be forced to appeal to some form of nonstationarity in the pair creation that results in the faster particles from a following bunch overtaking the slower particles in a preceding bunch ( usov 1987 ) . when the dispersive properties of waves in a pulsar plasma are considered in detail ( e.g. , lyutikov , blandford & machabeli 1999 ) ,
further difficulties arise . in the early literature ,
a possible analogy with growth of langmuir waves in solar radio bursts suggested that the growing waves might be longitudinal waves near @xmath52 in a relativistically - streaming , monoenergetic model .
however , the plasma is expected to have a large spread , @xmath53 , in lorentz factor ( e.g. , in the rest frame of the plasma ) . when such a spread is taken into account ,
parallel - propagating ( @xmath26 ) longitudinal waves have phase speeds @xmath51 only above a relatively high frequency @xmath54 in the plasma rest frame .
this frequency is further boosted , by the lorentz factor of the streaming motion , in the pulsar frame .
oblique ( l - o mode ) waves have phase speed @xmath51 only for a tiny range of angles about @xmath26 .
the dispersion curves in the rest frame are plotted in figure 2 for a small value of @xmath53 ; beam - driven growth of l - o mode waves is possible only for sufficiently small angles , @xmath55 , for which the dispersion curve extends into the region above the line @xmath56 in the figure .
these features lead to several unresolved problems ( melrose & gedalin 1999 ) .
first , the frequency of the beam - driven waves ( doppler boosted to the pulsar frame ) seems too high to account for all pulsar emission .
second , due to the curvature of the field lines , waves rapidly leave the tiny range of angles where growth is possible , severely restricting effective growth .
the suggestion that beam - driven growth may occur in other wave modes in the pulsar plasma leads to other difficulties : the x mode ( in a non - gyrotropic plasma ) does not couple directly to the beam ; alfvn waves can grow at lower frequencies , but can not escape directly due to a stop band , cf . ,
curve a in figure 2 .
the required large growth rate in a highly relativistic plasma precludes a maser mechanism : a maser operates in the random - phase regime , which requires that the growth rate be less than the bandwidth of the growing waves , which is very small in a highly relativistic plasma . a ( resonant ) reactive or hydrodynamic version of the instability applies for @xmath57 .
there is also a nonresonant reactive instability , which causes waves in a beam mode to grow at lower frequencies .
appeal to this nonresonant instability partly alleviates some of the difficulties ( gedalin , gruman & melrose 2002 ) .
_ other maser mechanisms _ : the other three instability - based mechanisms are linear acceleration emission ( melrose 1978 ; rowe 1995 ) , anomalous doppler instability ( machabeli & usov 1979 ) and curvature - drift instability ( kazbegi , machabeli & melikidze 1991 ) .
linear acceleration emission requires large - amplitude oscillations in the parallel electric field as the driving mechanism in the source region .
although there is no strong argument against this mechanism , there is no detailed model for the development of the required large - amplitude electrostatic oscillations and no independent evidence for them .
the other two mechanisms are possible only in the outer magnetosphere ( lyutikov , blandford & machabeli 1999 ) , and are sensitive to @xmath8 , both of which imply that they can not apply to all classes of pulsars .
also , both require that the refractive index be greater than unity at the point of emission , and this is even more difficult to satisfy than for the beam - driven cerenkov instability . _ orthogonal modes _ : there is compelling evidence that in at least some pulsars the emission emerges in two orthogonal modes ( oms ) that become spatially separated ( e.g. , mckinnon & stinebring 2000 ) .
the presence of oms requires refraction in a birefringent plasma that causes different rays for the two oms .
the observed oms raise a further difficulty : any instability favors one wave mode in the plasma over others , and after many growth times the faster growing mode completely dominates , implying emission in only one of the natural modes of the plasma .
hence , one requires some process that converts waves initially purely in one mode into a roughly equal mixture of two modes .
mode coupling due to large - scale gradients is one possibility , but a more effective process such as reflection off sharp boundaries seems to be necessary .
moreover , if such propagation effects are important , then the observed polarization should be determined at a polarization limiting region , rather than being characteristic of the emission mechanism or of the dominant mode at the source .
it has long been thought that the most plausible location for the source region of the radio emission is many tens of radii above the star in the polar cap regions .
recent analysis of data ( gangadhara & gupta 2001 ) favors a source region at such heights , roughly midway between the magnetic axis and the last closed field lines , far from the gap regions where pair creation is conventionally assumed to occur .
it may be that significant pair creation does occur between the conventional inner and outer gaps .
however , why should the radio emission originate from intermediate heights , rather than the gaps where the pair creation should be most efficient ? a speculation as to how some of these difficulties might be overcome involves appealing to a possible analogy with solar radio bursts .
type iii bursts are due to a stream of electrons , that generate langmuir waves through a cerenkov instability , and types i , ii and iii are all attributed to plasma emission .
theory implies that the polarization of plasma emission should be 100% in the o mode , and this is sometimes the case for type i bursts , but never the case for type ii and iii bursts , and some bursts of all types can be unpolarized .
a depolarizing agent operates , and this requires very small - scale inhomogeneities ( melrose 1975 ) , and may be associated with reflection off overdense structures , such as columns invoked to account for the directivity of type i emission ( bougeret & steinberg 1977 ) .
ducting is also required to account for the apparent sources being scatter images at much greater heights than the actual sources ( duncan 1979 ) .
the polarization , directivity and apparent height can all be explained in terms of a source in an underdense region with the escaping radiation reflected off nearly radial overdense structures .
nonstationary , inhomogeneous pair creation would produce dense columns of pair plasma that can act as ducts for radio emission generated in the underdense regions between the columns .
reflection off the columns can convert radiation in one mode into a mixture of the two modes , as well as ducting the radiation outward .
the radiation characteristic would then correspond to those of an apparent source at a much greater height than the actual source .
these speculations suggest that an appeal to locally highly time - variable pair creation may resolve some of the outstanding difficulties in the interpretation of the source location and polarization of pulsar radio emission .
high - energy emission processes are well understood , but the location of the source region is uncertain .
the radio mechanism is probably either relativistic plasma emission or torsion - driven curvature maser instability ; a detailed critical comparison of these is needed .
observed orthogonal mode polarization appears to require structures with sharp boundaries off which the waves can be reflected .
arendt , p.n . ,
jr & eilek , j.a .
2002 , , 581 , 451 blandford , r.d . 1975 , , 170 , 551 bougeret , j .- l . &
steinberg , j .- l .
1977 , , 61 , 777 canuto , v. , lodenquai , j. & ruderman , m. 1971 , , 3 , 2303 cheng , k.s . , ho , c. & ruderman , m.a . 1986 , , 300 , 522 cheng , k.s .
& zhang , l. 1999 , 515 , 337 daugherty , j.k . & harding , a.k .
1989 , 336 , 871 dermer . c.d .
1990 , , 360 , 214 duncan , r.a . 1979 , solarphys . , 63 , 389 gangadhara , r.t . &
gupta , y. 2001 , , 555 , 31 gedalin , m. , gruman , e. & melrose , d.b .
2002 , , 337 , 422 herold .
h. 1979 , 19 , 2868 hibschman , j.a . , & arons , j. 2001 , , 560 , 671 hirotani , k. & shibata , s. 2001 , , 325 , 1228 kazbegi , a.z . , machabeli , g.z . & melikidze , g.i . 1991 , , 263 , 377 kirk , j.g . , skraasen , o. , & gallant , y. 2002 , , 388 , l29 luo , q. & melrose , d.b .
1992 , , 258 , 616 luo , q. & melrose , d.b .
1995 , , 276 , 631 lyutikov , m. , blandford , r.d . & machabeli , g. 1999 , , 305 , 338 machabeli , g.z . &
usov , v.v .
1979 , sov.astron.lett .
, 5 , 445 mckinnon , m.m . & stinebring , d.r . 2000 , , 529 , 435 melrose , d.b .
1975 , solarphys .
, 43 , 79 melrose , d.b .
1978 , , 225 , 557 melrose , d.b . &
gedalin , m. 1999 , , 521 , 351 melrose . d.b . &
sy , w. 1972 , , 17 , 343 radhakrishnan , v. & deshpande , a.a . 2001 , , 379 , 551 romani , r. 1996 , , 470 , 469 rowe , e.t .
1995 , , 296 , 275 usov , v.v .
1987 , , 320 , 333 usov , v.v . &
melrose , d.b .
1996 , , 464 , 306 weise , j.i . & melrose , d.b .
2002 , , 329 , 115 zhang , b. & harding , a.k .
2000 , , 532 , 1150 | high - energy and radio emission mechanisms for pulsars are reviewed .
the source region for high energy emission remains uncertain .
two preferred radio emission mechanism are identified .
some difficulties may be resolved by appealing to nonstationary pair creation distributed widely in height . 1.25 in .125 in .25 in |
here we will go through some details of the matrix product state method , including the single photon state preparation and mps update of this state .
we will also present results of fidelity simulations not included in the main text .
the system - bath hamiltonian @xmath136 , where @xmath137 the operators @xmath138 , @xmath139 and @xmath140 are annihilation operators for photons in the cavity , phonons of mechanical system , and photons in the waveguide with mode index @xmath141 . note that in writing this hamiltonian , we are already in the rotating frame of the optical cavity frequency @xmath142 .
we can further go into the rotating frame with respect to the bath hamiltonian @xmath143 and the interaction term becomes @xmath144 where we have defined @xmath145 to calculate the evolution of the whole system , it is convenient to discretize time in small steps of @xmath45 . define the quantum noise increments in the time domain for the input field as : @xmath146 where @xmath147 and @xmath47 is an integer .
it is straightforward to verify that @xmath49 = \delta_{nm } $ ] from the commutation relation @xmath148 = \delta ( \omega - \omega^\prime ) $ ] of the field mode , which means all the time - bins can be interpreted as independent harmonic oscillators .
thus the hilbert space for the whole system is a tensor product space , including contributions from optomechanical system and quantum field in the waveguide , which can be modeled as a series of harmonic oscillators .
generally speaking , two things are required for the mps simulation : preparing the many - body state @xmath150 in mps form and updating this state step by step to evolve @xmath150 . in this and the next section
, we consider these two operations . as usual , at @xmath151 the state of the optomechanical system and the quantum field in the waveguide
are assumed to be completely uncorrelated and separable .
the state of the waveguide is composed of two temporal modes that are spaced apart from each other .
a large separation in time means that the state of the quantum field is very nearly the product state of two single photon states in two independent temporal modes .
a single photon in a temporal mode has a wavefunction given by @xmath152 .
this can be rewritten in the time - bin representation as _
n f_n _ = _ .
more generally , the joint state of the waveguide with two temporal modes each in their 0 or 1 state can be described by the linear combination of state vectors @xmath55 for @xmath56 or 1 . to put this state in an mps form , a sequence of singular value decompositions ( svd ) are required .
note that for separated temporal modes each being in the 0 or 1 photon states , the dimension of hilbert space for any of the time - bins can be chosen to be 2 .
in addition , we are operating in the polynomial @xmath153 subspace of the @xmath154 dimensional hilbert space of the waveguide only states @xmath155 are used since we ignore the possibility of two photons localizing at the same time - bin .
considering @xmath156 to be the dimension of the truncated hilbert space for the optomechanical system , the state of the full system would require @xmath157 parameters .
the mps representation takes advantage of this reduction in problem size in an implicit way .
in addition , we note that though we are limited to this subspace due to the photon number conserving symmetry of the hamiltonian and our selection of the initial state , our implementation of the mps method is capable of simulating the dynamics for a more complex input state in an efficient and automatic way , since the hilbert space is never explicitly reduced . 1 . for the first step , calculate the density matrix for the first time - bin as well as its eigenvalues and eigenvectors . use the eigenvalues to get the schmidt vector @xmath158 } $ ] and
expand the eigenvectors @xmath159}\rangle$ } } $ ] in the local basis \{@xmath160 } to get @xmath161 } $ ] , _ i.e. _ , @xmath162}\rangle$ } } = \gamma^{[1]i_1}_{\alpha_1 } $ ] .
2 . for the @xmath163th ( @xmath164 ) step , calculate the density matrix for the composite system of @xmath165 time - bins as well as its eigenvalues and eigenvectors . use the eigenvalues to get the schmidt vector @xmath166 } $ ] .
take the inner product @xmath167}|$ } } \langle i_k{\mbox{$|\psi_{\alpha_k}^{[k]}\rangle$ } } $ ] to give the element @xmath168i_k}_{\alpha_{k-1}\alpha_{k } } $ ] of @xmath168 } $ ] , where @xmath169 is the local basis \{@xmath160 } of the @xmath163th time - bin . to evolve the many body state , a sequence of unitary operators @xmath170
are applied to update the mps . here
@xmath171 as already mentioned in the main text , the mps used for simulation of our protocol keeps track of a waveguide of finite length equivalent to a delay time of @xmath61 and also the optomechanical system . 1 .
for @xmath47 from 1 to @xmath172 , apply @xmath173 to the optomechanical system and the @xmath47th time - bin . then use a swap gate to permute the order of them , so that @xmath174 can be applied in the same local way as @xmath60 . by the end of this step ,
the optomechanical system is at the last site of the mps .
2 . for @xmath47 from @xmath172 to 1 , perform swap gate to the optomechanical system and the @xmath47th time - bin .
note that in this step the many - body state is not evolved in time since it stays the same under swap gates .
repeat steps 1 and 2 to complete one repetition of the protocol
. the full evolution may be over multiple repetitions , @xmath63 .
the canonical mps form provides a convenient way to extract many properties of the system , like average values during evolution , correlation functions of the quantum field and so on .
in addition to the phase and fidelity that can be directly calculated from the wavefunction , we are particularly interested in the entanglement entropy of various bipartitions of the whole system .
these can be directly calculated from the @xmath166 } $ ] , the schmidt vector at time @xmath175 , in the mps representation .
we calculate @xmath1762}_\alpha \log_2 \lambda^{[k]2}_\alpha , \ ] ] which gives us the entanglement between part of the system , containing time - bins @xmath177 , and another part , with time - bins @xmath178 .
to simulate the effect of many reflections by the mirror , we use a series of swap gates to move the optomechanical system back every time it reaches the last site of the mps .
this scheme seems reasonable , and we justify it here in a more rigorous way .
the system as modelled is simply a closed system without losses . here
we wish to show formally the equivalence between our simulation scheme and the dynamics of the closed system . consider a long waveguide of length @xmath179 with linear dispersion , then @xmath180 where @xmath181 is the frequency of the fundamental mode and @xmath182 is the speed of light .
the summation starts from @xmath183 since we are in a rotating frame of cavity frequency . in this model
the delay time of the coherent feedback is @xmath184 . after going into the rotating frame of @xmath185 ,
the interaction term is @xmath186 where we have defined @xmath187 .
it is important to note that @xmath188 which is a direct result of the coherent feedback .
next let us discretize time in small steps of @xmath189 , where @xmath172 is a large integer and also the number of time - bins .
we define in the time domain @xmath190 , where @xmath191 is an integer and @xmath192 .
since @xmath193 for any @xmath191 , all the independent operators are @xmath194 .
as before , it is straightforward to show that @xmath195 = \delta_{ij } , \forall 1\leqslant i , j \leqslant n $ ] . therefore we have @xmath172 independent harmonic oscillators as time - bins . to describe the evolution of this state , we start again with the single - step evolution @xmath171 here @xmath196 is the operator defined in the time domain and @xmath197 .
this cyclic property of these unitary matrices evolving the state means that the total evolution operator @xmath198 over many steps can be written as @xmath199 for @xmath200 where @xmath201 are two integers and @xmath202 .
when @xmath203 goes from 1 to @xmath172 , the optomechanical system moves through the mps and interacts with the photons .
when @xmath203 becomes 1 again , a series of swap gates are required to move the optomechanical system back to the first site of the mps , so that @xmath204 can be applied on the mps in a local way .
this is exactly the simulation scheme we have implemented as described above .
the cutoff of the singular values in the singular value decomposition calculation is set to be @xmath205 . since the coupling between the cavity and the waveguide is the fastest process in the rotating frame ,
we choose @xmath206 .
the temporal width of the photon is chosen to be @xmath207 , _
i.e. _ , most of the photon is spread across @xmath208 time - bins to ensure that it can fully enter the cavity .
we choose the cutoff of the fock space to be @xmath209 for the optical cavity , @xmath210 for the mechanical oscillator and @xmath209 for each time bin .
the total waveguide is composed of @xmath211 time - bins , which is @xmath212 for @xmath108 .
the quantum optomechanical nonlinearity can cause the state of the quantum field in the waveguide to move outside of the subspace defined by the four state vectors @xmath98 , where there are @xmath213 photons in the temporal modes @xmath214 and @xmath26 of the waveguide .
there are three effects that cause this .
one is the change in the temporal mode after each bounce from the cavity .
the other is leakage due to the nonlinearity .
finally there is the effect of residual entanglement with the mechanical system after a repetition of the protocol .
we are interested in the latter two since the first can be in principle reversed with linear optical components .
we calculate the change in the overlaps @xmath215 starting from the initial state @xmath85 as defined in the main text , and with @xmath98 evolved separately with @xmath93 .
the results are plotted in figure [ fig : si_fig ] .
nominally , @xmath216 in the perfect case for every @xmath214 and @xmath26 , and so the infidelity is @xmath217 . | we show that the effective optical nonlinearity of a cavity optomechanical system can be used to implement quantum gates between propagating photons . by using quantum feedback
, we can enhance a slow and small optical nonlinearity to generate a large nonlinear phase shift between two spatially separated temporal modes of a propagating electromagnetic field .
this allows us to implement a gate between the two modes . after presenting a semiclassical derivation of the operation of the gate
, we verify the result by a full simulation of the state of the quantum field in the waveguide coupled to a cavity . to efficiently solve the schrdinger equation of the full system ,
we develop a matrix product state approach that keeps track of the entangled full quantum state of the coupled system .
these simulations verify the operation of the gate in the weak coupling regime where the semiclassical approximation is valid .
in addition , we observe a major reduction in gate fidelity as we approach the vacuum strong coupling regime of optomechanics where quantum fluctuations that are not accounted for in the semiclassical picture start to play a major role .
we consider possible experimental implementations of the gate with current technology .
the interaction between light and motion in cavity - optomechanical systems has enabled sensitive measurements of force and displacement , as well as quantum - optical control over the state of mechanical resonators @xcite .
conversely , an optically - coupled mechanical degree of freedom can lead to interactions between photons mediated by motion .
nonlinear optical effects such as wavelength conversion @xcite , generation of squeezed light @xcite , and electromagnetically induced transparency @xcite are manifestations of this optical nonlinearity and have been demonstrated in recent experiments .
nonetheless , a central goal of quantum optics and information is to generate nonlinearities that are large at the single - photon level @xcite .
it has been shown theoretically that it is possible to generate single - photon effective optical nonlinearities with optomechanics , but only with system parameters well outside of our current experimental reach @xcite .
the essential difficulty is that for a mechanically - induced optical nonlinearity to act on the incident light , the response of the mechanical system must be fast compared to the amount of time the photon spends inside the optical cavity .
moreover , the mechanical system should display a large response to the force induced by a single photon so that the cavity properties change significantly due to the presence of the photon requiring a highly compliant mechanical system .
these two contradicting needs , fast response and large compliance , are fundamentally at odds with each other and make strong effective photon - photon interactions extremely difficult to achieve in realistic cavity optomechanical systems .
in addition to the difficulty of generating the large kerr nonlinearities needed , it has been pointed out that in principle , subtle effects due to the multimode nature of a propagating light field make implementation of high - fidelity gates operating on flying photons using large instantaneous kerr nonlinearities impossible @xcite . here
we show that by introducing a _ coherent delay _ to a cavity - optomechanical system with a small and slow effective kerr nonlinearity , large photon - photon interactions between two spatially separated photons propagating in an optical fiber can be achieved .
moreover , we demonstrate using both a semiclassical argument and full quantum simulations taking into account the propagating quantum field with its many degrees of freedom , that a gate between two temporal modes of this field can be implemented with a cavity - optomechanical system that is in the bad - cavity regime .
the coherent delay allows a slow optomechanical system to induce a large effective interaction between the separated photons , greatly relaxing the optomechanical system requirements needed to implement such a gate .
the nonlocal nature of the gate also sidesteps a key assumption in the aforementioned impossibility arguments @xcite .
we consider an optomechanical system coupled to a long waveguide with an end mirror , so that photons can propagate back and forth inside the waveguide , and interact repeatedly with the mechanical resonator by entering the optical cavity .
these interactions and the slow internal dynamics of the mechanical system lead to an effective nonlinear interaction between the photons that is nonlocal and leads to a build - up of entanglement in the quantum fields inside the waveguide .
remarkably , quantum photonic operations can be implemented in such a scheme .
the protocol for the gate between two temporal modes of the light field is shown in figure [ fig : protocol ] .
the center frequencies of these two modes are chosen to be at the cavity frequency and each have a bandwidth much smaller than the cavity linewidth .
they are also separated by a time @xmath0 , where @xmath1 is the period of the mechanical oscillation and @xmath2 is the temporal extent of the photon .
our system is similar to and inspired by pulsed optomechanical experiments @xcite with the distinction that no measurement occurs in our feedback network
the photons are fed back to the system coherently .
the choice of delay time is essential for disentangling the mechanical system from the photons at the end of the protocol .
the mechanical system must return to its initial state regardless of the state of the input temporal modes , since any residual entanglement between the temporal modes and the mechanical system will reduce the fidelity of the gate .
as shown in figure [ fig : protocol ] and is justified below , this is the case for the temporal mode spacing we have chosen . .
the modes can be in states @xmath3 and @xmath4 .
( 1 ) the first photon mode interacts and is entangled with the mechanical oscillator since @xmath5 and @xmath6 will lead to different changes in the mechanical momentum and cause a state - dependent change in the mechanical state . after the mechanical system evolves for @xmath7 , ( 2 )
the second photon mode interacts with the mechanical oscillator and the mechanical system s state become entangled with both photons .
after the mechanical system evolves again for another @xmath8 , ( 3 ) the first photon temporal mode interacts again with the mechanical system causing its state to be disentangled from the mechanical oscillator .
after another @xmath8 of evolution , ( 4 ) the second photon mode comes back again and the mechanical system is decoupled from both temporal modes since it goes back to its ground state regardless of the initial states of the photon modes .
the blue lines signify the entanglement in the system and on the right the state - dependent evolution of mechanical state is shown in phase space.,scaledwidth=50.0% ] _ semiclassical model .
_ we consider a cavity - optomechanical system composed of an optical resonator at frequency @xmath9 with annihilation operator @xmath10 , coupled to a mechanical resonator with frequency @xmath11 and annhililation operator @xmath12 .
the waveguide coupling the the optomechanical system is terminated at the opposite end with a reflector .
the initial state of the waveguide consists of two temporal modes of the light field , each with extent @xmath13 .
we assume that @xmath14 so that each photon fully interacts with the cavity , exerting a significant impulse onto the mechanical system and obtaining a large uniform phase shift related to the internal state of the mechanical system , before propagating away in the waveguide .
each impulse is assumed to be nearly instantaneous on time scales relevant to the mechanical oscillator s motion , _
i.e. _ , @xmath15 . taken together , these conditions imply that we are operating in the bad cavity regime @xmath16 .
we can use a simplified semiclassical model to understand some of the behavior of this system .
the optomechanical system s hamiltonian is given by @xmath17 the radiation pressure force on the optomechanical system ( @xmath18 ) integrated over the interaction time of the photon with the cavity ( @xmath19 ) causes a rapid change in the momentum of the mechanical system by @xmath20 .
if we assume that the cavity can be eliminated adiabatically ( @xmath21 is large ) , the field in the cavity and waveguide may be related as @xmath22 , so after the interaction with the propagating photon , an impulse of @xmath23 is imparted onto the mechanical system .
we define a photon number operator @xmath24 , which counts the number of excitations in a window of size @xmath13 around the @xmath25 temporal mode ( @xmath13 and @xmath26 only set the limits of integration over the field operators ) .
the interaction between the @xmath25 temporal mode and the mechanical system is given by the unitary operator & = & ( -i/ ) + & = & ( -4i g_0 /x _ ) , with position @xmath27 .
the mechanical free evolution operator is @xmath28 .
our protocol can then be compactly stated as @xmath29 , with = .
this sequence of interactions is shown schematically in figure [ fig : protocol ] .
we can now calculate the result of these operations on the joint state of the optomechanical systems and photonic waveguide , given that the waveguide s photonic states are initialized as @xmath30 , @xmath31 , @xmath32 , and @xmath33 , and the mechanical mode is in its ground state .
for the first three of these states , @xmath34 , while for the last one , @xmath35 .
this demonstrates that the internal dynamics of the optomechanical system executes a controlled - phase gate on to the joint state of the two photons .
following this calculation through , we find the phase shift imparted on the @xmath33 state after one run of the protocol is given by _ 1 = 32 ( g_0/)^2 . according to this simplified model , obtaining a total phase shift of @xmath36
can be accomplished either by going to a large enough coupling @xmath37 , or by running the protocol multiple times , @xmath38 , for a smaller @xmath37 so that a total phase shift builds up over several bounces . to distinguish between these two approaches , it is important to realize that in any real system , in addition to the coupling between the cavity and the waveguide , there are other fixed intrinsic optical loss channels characterized by a loss rate @xmath39 leading to a probability @xmath40 that a photon is lost on each bounce . according to the semiclassical model ,
since the chance of photon loss in multiple runs is proportional to @xmath41 , and @xmath42 , it is always advantageous to make @xmath21 as small as possible .
this conclusion however neglects another source of decoherence which arises due to quantum fluctuations .
as we will see below , the semiclassical model fails to capture some of the essential physics of the problem . at the most basic level
, the model assumes that the cavity field follows adiabatically the input field from the waveguide , since they are resonant and the photon has a spectrum that is much narrower than that of the cavity .
however , at large @xmath37 , the quantum motion of the mechanical system imposes a jitter on the optical cavity frequency that is comparable to the optical cavity linewidth @xcite .
this extra noise prevents the protocol from operating as intended .
in fact , the final state of the system will exhibit significant entanglement between the photons and mechanical resonator preventing the protocol from functioning as a photon - photon gate . as described below , smaller @xmath37
are required , leading to a trade - off between the quantum - fluctuation - induced loss in fidelity , and the effect of intrinsic optical losses .
for a quantitative understanding of this trade - off we consider a full quantum model of the system . _
quantum dynamics .
_ most theoretical methods of quantum optics are best suited for modeling coupling between a system and a reservoir that is markovian in nature .
the protocol we are considering utilizes a waveguide that reflects the light back onto the system realizing a quantum feedback network and dynamics that are highly non - markovian .
many methods for understanding feedback networks have been developed in recent years @xcite , applied to optomechanical systems @xcite , and experimentally realized @xcite .
the feedback network we are considering requires a long delay due to the slow dynamics of the mechanical oscillator , and recently methods for understanding dynamics of systems in such quantum feedback networks have been proposed @xcite . in addition , we are interested in understanding the evolution of the state of the photons in the waveguide over many bounces , so tracing out the bath would be akin to throwing the photons out with the bath degrees of freedom .
one approach @xcite to solving the full dynamics of the waveguide and system together is to discretize the waveguide into time steps @xmath43 that are much smaller than any of the relevant dynamics of the system and evolve what is now effectively the interaction between a 1d chain of harmonic oscillators and an optomechanical system . in principle , keeping track of the state of such a 1d chain is daunting due to the exponentially large hilbert space , where @xmath44 is the truncated dimension of each bin s fock space . ]
required for its full representation .
luckily , the states of the waveguide we are considering have far less entanglement than general states in the full hilbert space , so efficient methods for storing and evolving the states can be utilized @xcite .
our full quantum simulation is based on the matrix product state ( mps ) representation of the quantum field .
mps as well as the related time - dependent density matrix renormalization group ( tdmrg ) techniques have already been well - established in condensed matter physics for simulating 1d quantum many - body systems .
the success of these methods comes from their efficient description of quantum system with moderate amounts of entanglement @xcite .
it was realized that matrix product states are also closely related to entangled fields at the output of photonic systems @xcite .
more recently , pichler and zoller @xcite implemented an mps - based simulation of photonic delayed quantum feedback networks . here
we will use a method similar to that presented in ref .
@xcite and outlined in the appendix . to implement mps method for the continuous quantum field of the waveguide , we discretize time in small steps @xmath45 and define operator @xmath46 for the @xmath47th time - bin , where @xmath48 and @xmath47 is an integer .
it is straightforward to verify that @xmath49 = \delta_{nm } $ ] which means the time - bins can be interpreted as independent harmonic oscillators .
the state of this quantum many body system can be represented in the canonical mps form @xcite .
a spatially distributed single photon state of the waveguide is then @xmath50 , where @xmath51 .
we consider an initial state where the optomechanical system s optical and mechanical modes are both in their ground states , and consider two temporal modes of the photon with annihilation operators @xmath52 and @xmath53 separated by a time interval @xmath54 .
we identify the states @xmath55 for @xmath56 or 1 are to an extremely good approximation orthogonal given a temporal separation that is much larger than their width .
] , and assume an initial state of the whole system _ & = & ( 1 + ) ( 1 + ) _ _ _ /2,[eqn : initstate ] + & = & ( + + + ) .
initially , the optomechanical system is at the first site of the mps . to evolve the many - body state , we sequentially update the mps by applying the unitary @xmath57 as a local gate on the @xmath58 time - bin and the optomechanical system .
then a swap gate is used to permute the order of them so that @xmath59 can be applied in the same way as @xmath60 .
this process effectively simulates a discrete representation of the quantum stochastic schrdinger equation ( qsse ) @xcite and is continued until the optomechanical system reaches the last site of the mps .
in contrast to ref .
@xcite where the number of sites in the mps is proportional to the total simulation time , here our waveguide and therefore our mps representation are of finite length due to the feedback .
after every @xmath61 time interval , the system has interacted with the whole waveguide and reaches the last site of the mps .
we simulate the reflection due to the far mirror by moving the optomechanical system through the mps back to the first site via a series of swap gates . at this point
, the evolution can be continued as before to simulate interaction after one bounce .
this process is repeated @xmath62 times for @xmath63 runs of the protocol .
a more detailed description and justification can be found in the appendix . of the fully coupled waveguide - optomechanical system for @xmath64 .
( b ) evolution of the entangelment entropy @xmath65 between the two localized photons over many repetitions of the protocol .
the first four points are the same as those in ( a ) .
( c ) simulated phase shift @xmath66 after @xmath67 .
the dashed line is comparison with the semiclassical @xmath68 .
the mps simulations in ( a ) and ( b ) were done for @xmath69 , @xmath70 , while in ( c ) @xmath37 was varied .
( d ) the simulated phase shift over multiple repetitions of the protocol for @xmath69 is plotted .
the circles use @xmath65 to infer @xmath71 , while the black trace is calculated directly by taking the argument of @xmath72 .
there is good agreement between the two and the phase increases linearly as @xmath73 .
, scaledwidth=45.0% ] in the mps representation we decompose the state vector of the @xmath74-site system , @xmath75 , into a product of tensors @xmath76i_n}_{\alpha_{k-1}\alpha_{k}}$ ] and vectors @xmath77}_{\alpha_k}$ ] for @xmath78 , such that a bipartition of the state at bond @xmath79 can be written as a schmidt decomposition @xmath80}_{\alpha_l } { \mbox{$|\phi^{[1\ldots l]}_{\alpha_l}\rangle$}}{\mbox{$|\phi^{[(l+1)\ldots n]}_{\alpha_l}\rangle$}}$ ] .
this allows us to calculate the entanglement entropy between the two halves of the system by simply reading off the values of the @xmath81 vector at bond @xmath79 , and calculating @xmath822}_{\alpha_l } \log_2 \lambda^{[l]2}_{\alpha_l}$ ] . the black line in figure [ fig : result1](a )
is a plot of the @xmath83 as a function of time @xmath84 for the initial state @xmath85 .
the two peaks at the positions of the two photons correspond to the finite width photon excitations in the waveguide , since one excitation is spread across multiple local sites and detecting the photon at a particular bin tells us that there are no photons in the nearby bins .
stated more concisely , @xmath86 can not be written as a product state of time - bin localized photon excitations .
in addition to the localized peaks , two other interesting regions are that between the photons , and the region between the photons and the optomechanical system .
we call the amount of entanglement entropy between the photons and mechanical resonator @xmath87 and call the entanglement entropy between the two photons @xmath65 .
as expected , both @xmath87 and @xmath65 are zero since the initial state can be written as a product state of separated temporal modes of the optical field in the waveguide and the optomechanical system , c.f .
equation ( [ eqn : initstate ] ) .
the entanglement entropy throughout the waveguide is plotted after a few subsequent runs of the protocol and shown in dashed lines in figure [ fig : result1](a ) . for the system parameters in figure [ fig : result1](a ) , @xmath65 increases on each run of the protocol while @xmath87 remains close to zero .
the evolution of @xmath65 for more runs of the protocol is plotted in fig . [
fig : result1](b ) and shows clear oscillatory behavior .
this can be attributed to the evolving phase shift of the @xmath88 state with respect to the other states , since for a state @xmath89 the entanglement entropy is @xmath90/2 $ ] , where @xmath91 . to verify that this is in fact the correct interpretation
, we calculate the phase shift directly from the wavefunctions by taking the overlap between a vector @xmath92 evolved on a system with @xmath93 , and the result of our simulation with larger @xmath94 .
we compare this phase to the semiclassically predicted results @xmath95 , and find good agreement at smaller @xmath37 as shown in figure [ fig : result1](c ) .
the agreement becomes progressively worse at higher coupling .
the total phase shift @xmath96 increases linearly as @xmath97 in fig . [ fig : result1](d ) , which explains the oscillatory behavior of entanglement entropy .
in addition to calculating the phase , we must verify that the state of the electromagnetic field after interaction with the optomechanical system remains within the subspace of states @xmath98 , modulo the linear optical response of cavity .
we calculate the fidelity @xmath99 after one and two bounces of the photons from the cavity .
nominally , after two bounces , _
i.e. _ , @xmath67 , we expect @xmath100 for the input state in equation ( [ eqn : initstate ] ) . in figure
[ fig : result2](a ) a plot of the infidelity @xmath101 against the interaction rate is shown for one and two bounces of the waveguide photons from the cavity .
after one bounce , the significant entanglement between the mechanical system and waveguide photons leads to higher infidelity .
after two bounces , _
i.e. _ , a full run of the protocol , the entanglement between the mechanics and waveguide photons is largely erased , causing an increase in the overlap @xmath102 .
larger interactions cause a breakdown of this picture and lead to residual entanglement with the mechanical system and lower fidelity . for @xmath103 , we simulate a fidelity of @xmath104 after two bounces , and @xmath105 after the @xmath106 required to implement the gate with @xmath63 is shown in the supplementary materials . ] .
in other words the photon wavefunction is not significantly distorted in the interaction , as would be expected for large instantaneous kerr nonlinearities @xcite .
this is clear evidence that a gate between the two spatially separated photons can been implemented with this protocol . as a function of coupling strength @xmath37 .
the infidelity is very low after two bounces and increases with coupling strength . after only one bounce ,
the infidelity is much larger , since the mechanical system is no longer in its ground state .
( b ) @xmath87 between photons and the mechanical system for different values of @xmath107 and @xmath63 . here
@xmath108 .
a transition to the quantum regime is visible at @xmath109 , leading to large entanglement between the mechanical system and the waveguide photon.,scaledwidth=45.0% ] another signature of the failure of the gate is the presence of residual entanglement between the photons and the mechanical system after running the protocol , _
i.e. _ , a non - zero value for @xmath87 . in fig .
[ fig : result2](b ) , the evolution of @xmath87 is plotted against @xmath37 and @xmath63 of the protocol .
an interesting _ threshold _ feature is visible at a critical value of @xmath110 .
this establishes a clear breakdown of the semiclassical regime , and a boundary between two different regimes , with the semiclassical regime being more relevant for the gate .
we note that this critical value of @xmath37 occurs close to the value where the displacement due to the momentum imparted by a single photon , @xmath111 becomes equal with the zero - point fluctuations in the mechanical motion @xmath112 .
_ dissipation and experimental prospects .
_ we have shown that strong photon - photon interactions can be obtained with an optomechanical system that is outside the strong coupling regime ( @xmath113 ) .
we estimate the effect of five sources of decoherence ; ( 1 ) the mechanical coupling to the thermal bath , ( 2 ) the effect of other weakly - coupled mechanical modes and sources of phase fluctuations , ( 3 ) the intrinsic optical loss @xmath39 , ( 4 ) insertion loss , and ( 5 ) the propagation losses in the long waveguide .
the important parameter in understanding the mechanical decoherence ( 1 ) is the thermalization rate @xmath114 , where @xmath115 is the mechanical linewidth and @xmath116 is the thermal occupation at the mechanical frequency . in the high temperature limit , @xmath117 and @xmath118
as long as the @xmath119 , the chance of a phonon entering the system from the bath during the time that protocol is being run for remains small .
this is equivalent to having @xmath120 .
such @xmath121 products have been obtained at cryogenic tempeatures @xcite , and more recently at room temperature with silicon nitride membranes @xcite .
the effect of competing mechanical modes ( 2 ) has long been an issue in pulsed optomechanics experiments @xcite .
recent experiments @xcite have been successful at reducing the coupling to these modes to a few percent of the primary mode by careful engineering of structures and positioning of the optical beam . sources ( 3)-(5 ) affect the protocol in a similar way , in the sense that they introduce a finite probability @xmath122 that a photon can be lost to the environment during the execution of the gate .
for the intrinsic losses ( 3 ) , this means requiring @xmath123 is composed of the intrinsic and extrinsic parts @xmath124 , with @xmath125 . ]
be close to @xmath126 .
in other words , the coupling ratio @xmath127 . since @xmath128 , it is best to reduce @xmath21 to smallest value allowable by the full quantum model .
this leads to @xmath106 , meaning that @xmath129 for a @xmath130 . for insertion losses ( 4 )
, it is important to keep @xmath131 also small for similar reasons . finally (
5 ) , the chance of a photon being absorbed in the long delay is related to the loss rate of the fiber ( on the order of @xmath132 in the telecom c - band ) and the required delay @xmath133 . to achieve a more than @xmath134 chance for the photon to survive propagation in the delay for the @xmath106 case shown in figure [ fig : result1 ] , we require a mechanical frequency greater than @xmath135 .
currently , these parameters remain outside our experimental reach , though progress in experimental quantum optomechanics is expected to bring us closer to the required parameters in a few platforms in the coming years . in conclusion , we have demonstrated that a gate between propagating photons can be implemented with the slow and weak nonlinearity of an optomechanical resonator under the correct quantum feedback conditions .
in addition to opening up a new part of the optomechanical parameter space to quantum experiments , the approach shows that quantum feedback has the potential to enhance quantum nonlinear phenomena . in the future , it is interesting to consider the evolution of entanglement in quantum feedback networks using the mps method , to better understand how interesting quantum many - body states with nontrivial correlations can be generated .
* acknowledgements .
* this work was supported by nsf eccs-1509107 and the stanford terman fellowship , tshinghua university undergraduate research program , onr muri qomand , as well as start - up funds from stanford university .
we thank jeff hill for his comments on the manuscript .
46ifxundefined [ 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/revmodphys.86.1391 [ * * , ( ) ] link:\doibase 10.1126/science.1228370 [ * * , ( ) ] , http://dx.doi.org/10.1038/ncomms2201 http://www.nature.com/ncomms/journal/v3/n11/suppinfo/ncomms2201{_}s1.html [ * * , ( ) ] link:\doibase 10.1038/nphys2911 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.116.043601 [ * * , ( ) ] , link:\doibase 10.1038/nature11325 [ * * , ( ) ] , link:\doibase 10.1038/nature12307 [ * * , ( ) ] link:\doibase 10.1103/physrevx.3.031012 [ * * , ( ) ] , link:\doibase 10.1126/science.1195596 [ * * , ( ) ] http://arxiv.org/abs/1012.1934 [ * * , ( ) ] link:\doibase 10.1103/physreva.52.3489 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.107.063601 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.063602 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.063601 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.013603 [ * * , ( ) ] , link:\doibase 10.1103/physreva.73.062305 [ * * , ( ) ] link:\doibase 10.1103/physreva.81.043823 [ * * , ( ) ] , link:\doibase 10.1103/physreva.89.052327 [ * * , ( ) ] link:\doibase 10.1073/pnas.1105098108 [ * * , ( ) ] link:\doibase 10.1038/ncomms3295 [ * * , ( ) ] link:\doibase 10.1109/tac.2009.2031205 [ * * , ( ) ] , link:\doibase 10.1088/1367 - 2630/16/6/065004 [ * * , ( ) ] , link:\doibase 10.1103/physrevapplied.1.054005 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.173602 [ * * , ( ) ] , link:\doibase 10.1103/physreva.87.013815 [ * * , ( ) ] , link:\doibase 10.1103/physrevx.3.021013 [ * * , ( ) ] , link:\doibase 10.1364/oe.21.018371 [ * * , ( ) ] , @noop ph.d .
thesis , ( ) link:\doibase 10.1103/physrevlett.115.060402 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.116.093601 [ * * , ( ) ] , link:\doibase 10.1140/epjqt / s40507 - 016 - 0041 - 9 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.91.147902 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.93.040502 [ * * , ( ) ] , link:\doibase 10.1016/j.aop.2010.09.012 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.95.110503 [ * * , ( ) ] , link:\doibase 10.1103/physreva.75.032311 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.104.190405 [ * * , ( ) ] , link:\doibase 10.1038/nature10461 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.116.147202 [ * * , ( ) ] , link:\doibase 10.1103/physrevx.6.021001 [ * * , ( ) ] , http://arxiv.org/abs/1608.00937 [ ( ) ] , |
this document contains a derivation of the likelihood - ratio formula for continuous quantum measurements with poissonian noise in sec .
[ poisson ] and an example of quantum optomechanical stochastic force detection in sec . [ force ] .
[ [ poissonlikelihood - ratio - formula - for - continuous - poissonian - measurements ] ] [ poisson]likelihood - ratio formula for continuous poissonian measurements ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the completely positive map for a weak poissonian measurement is given by @xmath103}\right\}},\end{aligned}\ ] ] where @xmath104 and @xmath105 models the effect of imperfect quantum efficiency @xmath64 . rearranging terms @xcite , @xmath106 } , \\
\tilde p(\delta y ) & \equiv ( 1-\delta y)(1-\alpha\delta t)+ \delta y\alpha\delta t,\end{aligned}\ ] ] where @xmath43 is a reference probability distribution and @xmath65 is an arbitrary positive number
. this gives @xmath107 equation ( [ df ] ) coincides with the quantum filtering equation for an unnormalized posterior density operator @xmath46 given poissonian observations @xmath57 @xcite .
next , consider @xmath108 where @xmath66 is the filtering estimate of the observable @xmath109 assuming that the hypothesis @xmath33 is true .
expanding @xmath110 in taylor series , @xmath111 with @xmath112 for @xmath113 and @xmath114 for poissonian observations , @xmath115}.\end{aligned}\ ] ] [ [ force - quantum - optomechanical - detection - of - a - gaussian - stochastic - force ] ] [ force ] quantum optomechanical detection of a gaussian stochastic force ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ let @xmath116 be a classical force acting on a moving mirror with position operator @xmath84 and momentum @xmath85 , and @xmath68 be a vectoral classical gaussian stochastic process @xmath68 described by the ito equation @xmath117 the mirror is assumed to be a harmonic oscillator with mass @xmath118 and frequency @xmath119 and part of an optical cavity pumped by a near - resonant continuous - wave laser .
the phase quadrature of the cavity output is measured continuously by homodyne detection , with an observation process given by @xmath57 . for simplicity ,
i assume that the optical intracavity dynamics can be adiabatically eliminated , the phase modulation by the mirror motion is much smaller than @xmath120 radians , such that the homodyne detection is effectively measuring the mirror position , and there is no excess decoherence . under hypothesis @xmath1 ,
the force is present and the quantum filtering equation for the unnormalized hybrid density operator @xmath121 is then given by @xcite @xmath122 + \mathcal l_c(x ) f_1 , \\
h_1(x ) & \equiv \frac{p^2}{2 m } + \frac{m\omega^2}{2}q^2-q cx , \\ \mathcal l_c(x )
f_1 & \equiv -\sum_\mu { \frac{\partial } { \partial x_\mu}}{\left[(a x)_\mu f_1\right ] } + \frac{1}{2}\sum_{\mu,\nu } { \frac{\partial ^2}{\partial x_\mu\partial x_\nu } } { \left(b_{\mu\nu}f_1\right)},\end{aligned}\ ] ] where @xmath42 is the measurement noise variance that depends on the laser intensity and the cavity properties and @xmath123 is the forward kolmogorov generator for the classical process @xmath68 . under the null hypothesis @xmath0 ,
the force is absent and the filtering equation for the oscillator density operator @xmath81 is @xmath124 , \\
h_0 & \equiv \frac{p^2}{2 m } + \frac{m\omega^2}{2}q^2.\end{aligned}\ ] ]
these filtering equations can be transformed to equations for the wigner functions of @xmath46 : @xmath125 } , \\ \mathcal l_0 ' g_0 & \equiv -\frac{p}{m}{\frac{\partial g_0}{\partial q}}+m\omega^2 q{\frac{\partial g_0}{\partial p } } + \frac{\hbar^2}{8r}{\frac{\partial^2 g_0}{\partial p^2}},\end{aligned}\ ] ] where @xmath84 and @xmath85 are now phase - space variables ,
@xmath85 is seen to suffer from measurement - back - action - induced diffusion , and @xmath126 has the form of a forward kolmogorov generator for a new gaussian process @xmath127 : @xmath128 } + \frac{1}{2}\sum_{\mu,\nu } { \frac{\partial ^2}{\partial z_\mu\partial z_\nu } } { \left(s_{1\mu\nu}g_1\right ) } , \\
j_1 & \equiv { \left(\begin{array}{cc|c}0 & 1/m & { \boldsymbol{0 } } \\ -m\omega^2 & 0 & c\\ \hline { \boldsymbol{0 } } & { \boldsymbol{0 } } & a \end{array}\right ) } , \quad s_1 \equiv { \left(\begin{array}{cc|c}0 & 0 & { \boldsymbol{0 } } \\ 0 & \hbar^2/4r & { \boldsymbol{0}}\\ \hline { \boldsymbol{0 } } & { \boldsymbol{0 } } & b \end{array}\right)},\end{aligned}\ ] ] with @xmath129 denoting zero matrices .
similarly , under @xmath0 , we have @xmath130 and @xmath131 } + \frac{1}{2}\sum_{\mu,\nu } { \frac{\partial ^2}{\partial z_\mu\partial z_\nu } } { \left(s_{0\mu\nu}g_0\right ) } , \\
j_0 & \equiv { \left(\begin{array}{ccc}0 & 1/m \\ -m\omega^2 & 0 \end{array}\right ) } , \quad s_0 \equiv { \left(\begin{array}{ccc}0 & 0 \\ 0 & \hbar^2/4r \end{array}\right)}.\end{aligned}\ ] ] the gaussian statistics mean that we can use kalman - bucy filters to compute the filtering estimates of the mirror position @xmath66 given @xmath57 @xcite : @xmath132 and the likelihood ratio becomes @xmath133}.\end{aligned}\ ] ] given the gaussian structure of the problem under each hypothesis , we can use known results about the chernoff upper bounds for classical waveform estimation to bound the error probabilities @xcite : @xmath134 } , \\
p_{01 } & \le \exp{\left[\mu(s)+(1-s)\gamma\right ] } , \\
\mu(s ) & = \frac{1}{2r}\int_{t_0}^t dt { \left[(1-s)\sigma_{1q}(t)+s\sigma_{0q}(t ) -\tilde\sigma_{q}(s , t)\right]},\end{aligned}\ ] ] where @xmath135 , @xmath8 is the threshold of the likelihood - ratio test , @xmath136 is the @xmath84 variance component of @xmath137 , which obeys eq .
( [ sigma ] ) , and @xmath138 is the variance of @xmath139 for a different filtering problem , in which observations of @xmath139 are made with noise variance @xmath42 and @xmath140 has the statistics of @xmath84 under @xmath33 , viz .
, @xmath141 the tightest upper bounds are obtained by minimizing the bounds with respect to @xmath142 .
if @xmath68 and therefore @xmath84 are stationary , @xmath137 and @xmath143 will converge to steady states in the long - time limit , and the chernoff bounds will decay exponentially with time . | i propose a general quantum hypothesis testing theory that enables one to test hypotheses about any aspect of a physical system , including its dynamics , based on a series of observations .
for example , the hypotheses can be about the presence of a weak classical signal continuously coupled to a quantum sensor , or about competing quantum or classical models of the dynamics of a system .
this generalization makes the theory useful for quantum detection and experimental tests of quantum mechanics in general . in the case of continuous measurements ,
the theory is significantly simplified to produce compact formulae for the likelihood ratio , the central quantity in statistical hypothesis testing .
the likelihood ratio can then be computed efficiently in many cases of interest .
two potential applications of the theory , namely quantum detection of a classical stochastic waveform and test of harmonic - oscillator energy quantization , are discussed .
testing hypotheses about a physical system by observation is a fundamental endeavor in scientific research .
observations are often indirect , noisy , and limited ; to choose the best model of a system among potential candidates , statistical inference is the most logical way @xcite and has been extensively employed in diverse fields of science and engineering .
many important quantum mechanics experiments , such as tests of quantum mechanics @xcite , quantum detection of weak forces or magnetic fields @xcite , and quantum target detection @xcite , are examples of hypothesis testing . to test quantum nonlocality , for instance ,
one should compare the quantum model with the best classical model ; bell s inequality and its variations , which impose general bounds on observations of local - hidden - variable systems , have been widely used in this regard @xcite .
the analyses of experimental data in many such tests have nonetheless been criticized by peres @xcite : the statistical averages in all these inequalities can never be measured exactly in a finite number of trials .
one should use statistical inference to account for the uncertainties and provide an operational meaning to the data .
another important recent development in quantum physics is the experimental demonstration of quantum behavior in increasingly macroscopic systems , such as mechanical oscillators @xcite and microwave resonators @xcite . to test the quantization of the oscillator energy @xcite , for example
, the use of quantum filtering theory has been proposed to process the data @xcite , but testing quantum behavior by assuming quantum mechanics can be criticized as begging the question .
an ingenious proposal by clerk _
et al . _
considers the third moment of energy as a test of energy quantization @xcite . like the correlations in bell s inequality ,
however , the third moment is a statistical average and can not be measured exactly in finite time .
again , statistical inference should be used to test the quantum behavior of a system rigorously , especially when the measurements are weak and noisy .
the good news here is that the error probabilities for hypothesis testing should decrease exponentially with the number of measurements when the number is large @xcite , so one can always compensate for a weak signal - to - noise ratio by increasing the number of trials .
quantum hypothesis testing was first studied by holevo @xcite and yuen _ et al . _
@xcite .
since then , researchers have focused on the use of statistical hypothesis testing techniques for initial quantum state discrimination @xcite . here
i propose a more general quantum theory of hypothesis testing for _ model _ discrimination , allowing the hypotheses to be not just about the initial state but also about the dynamics of the system under a series of observations .
this generalization makes the theory applicable to virtually any hypothesis testing problem that involves quantum mechanics , including tests of quantum dynamics @xcite and quantum waveform detection @xcite . in the case of continuous measurements with gaussian or poissonian noise , the theory
is significantly simplified to produce compact formulae for the likelihood ratio , the central quantity in statistical hypothesis testing .
the formulae enable one to compute the ratio efficiently in many cases of interest and should be useful for numerical approximations in general .
notable prior work on continuous quantum hypothesis testing is reported in refs .
@xcite , which study state discrimination or parameter estimation only and have not derived the general likelihood - ratio formulae proposed here .
to illustrate the theory , i discuss two potential applications , namely quantum detection of a classical stochastic waveform and test of harmonic - oscillator energy quantization .
waveform detection is a basic operation in future quantum sensing applications , such as gravitational - wave detection , optomechanical force detection , and atomic magnetometry @xcite .
tests of energy quantization , on the other hand , have become increasingly popular in experimental physics due to the rapid recent progress in device fabrication technologies @xcite . besides these two applications , the theory is expected to find wide use in quantum information processing and quantum physics in general , whenever new claims about a quantum system need to be tested rigorously .
statistical hypothesis testing entails the comparison of observation probabilities conditioned on different hypotheses @xcite . to test two hypotheses
labeled @xmath0 and @xmath1 using an observation record @xmath2 , the observer splits the observation space into two parts @xmath3 and @xmath4 ; when @xmath2 falls in @xmath3 , the observer chooses @xmath0 , and when @xmath2 falls in @xmath4 , the observer chooses @xmath1 .
the error probabilities are then @xmath5 and @xmath6 .
all binary hypothesis testing protocols involve the computation of the likelihood ratio , defined as @xmath7 the ratio is then compared against a threshold @xmath8 that depends on the protocol ; one decides on @xmath1 if @xmath9 and @xmath0 if @xmath10 .
for example , the neyman - pearson criterion minimizes @xmath11 under a constraint on @xmath12 , while the bayes criterion minimizes @xmath13 , @xmath14 and @xmath15 being arbitrary positive numbers . for multiple independent trials ,
the final likelihood ratio is simply the product of the ratios . in most cases ,
the error probabilities are difficult to calculate analytically and only bounds , such as the chernoff upper bound @xcite , may be available , but the likelihood ratio can be used to update the posterior hypothesis probabilities from prior probabilities @xmath16 and @xmath17 via @xmath18 and @xmath19 and therefore quantifies the strength of evidence for @xmath1 against @xmath0 given @xmath2 @xcite .
generalization to multiple hypotheses beyond two is also possible by computing multiple likelihood ratios or the posterior probabilities @xmath20 @xcite .
consider now two hypotheses about a system under a sequence of measurements , with results @xmath21 .
for generality , i use quantum theory to derive @xmath22 for both hypotheses , but note that a classical model can always be expressed mathematically as a special case of a quantum model .
the observation probability distribution is @xcite @xmath23,\end{aligned}\ ] ] where @xmath24 is the initial density operator at time @xmath25 , @xmath26 is the completely positive map that models the system dynamics from time @xmath27 to @xmath28 , @xmath29 is the completely positive map that models the measurement at time @xmath28 , and the subscripts @xmath30 for @xmath24 , @xmath31 , and @xmath32 denote the assumption of @xmath33 for these quantities . to proceed , let @xmath34 and assume the following kraus form of @xmath32 for gaussian measurements
@xcite : @xmath35 } \nonumber\\&\quad\times m_j(\delta z , t)\rho m_j^\dagger(\delta z , t ) , \nonumber\\ m_j(\delta z , t ) & \equiv \frac{1}{(2\pi q_j\delta t)^{1/4 } } \exp{\left(-\frac{\delta z^2}{4q_j\delta t}\right ) } \nonumber\\&\quad\times { \left[1+\frac{\delta z}{2q_j}c_j-\frac{\delta
t}{8q_j}c_j^\dagger c_j + o(\delta t)\right]},\end{aligned}\ ] ] where @xmath36 is the noise variance of the inherent quantum - limited measurement , @xmath37 is the excess noise variance , @xmath38 is a quantum operator depending on the measurement , and @xmath39 denotes terms asymptotically smaller than @xmath40 .
the map becomes @xmath41 , \nonumber\\ \tilde p(\delta y ) & \equiv \frac{1}{\sqrt{2\pi r\delta t } } \exp{\left(-\frac{\delta y^2}{2r\delta t}\right ) } , \quad r \equiv q_j + s_j.\end{aligned}\ ] ] i assume that the total noise variance @xmath42 is independent of the hypothesis to focus on tests of hidden models rather than the observation noise levels .
@xmath43 then factors out of both the numerator and denominator of the likelihood ratio and cancels itself . taking the continuous time limit using it calculus with @xmath44 ,
the likelihood ratio becomes @xmath45 , with @xmath46 obeying the following stochastic differential equation : @xmath47 and @xmath48 being the lindblad generator originating from @xmath49 . equation ( [ dmz ] ) has the exact same mathematical form as the linear belavkin equation for an unnormalized filtering density operator @xcite , but beware that @xmath46 represents the state of the system only if @xmath33 is true ; i call @xmath46 an assumptive state . to put @xmath50 in a form more amenable to numerics , consider the stochastic differential equation for @xmath51 : @xmath52 where @xmath53 is an assumptive estimate ; it is the posterior mean of the observable @xmath54 only if @xmath33 is true .
the form of eq .
( [ dtrf ] ) suggests that it can be solved by taking the logarithm of @xmath51 , i.e. , @xmath55 , resulting in @xmath56 with the @xmath57 integral being an it integral .
@xmath50 becomes @xmath58}. \label{formula}\end{aligned}\ ] ] this compact formula for the likelihood ratio is the quantum generalization of a similar result by duncan and kailath in classical detection theory @xcite .
generalization to the case of vectoral observations with noise covariance matrix @xmath42 is trivial ; the result is simply eq .
( [ formula ] ) with @xmath59 replaced by @xmath60 and @xmath61 by @xmath62 . for continuous measurements with poissonian noise ,
a formula for @xmath50 can be derived similarly @xcite : @xmath63 } , \label{formula2 } \\ \mu_j & \equiv \frac{1}{{\operatorname{tr}}f_j}{\operatorname{tr}}(\eta_j c_j^\dagger c_j f_j ) , \label{estimate2 } \\
df_j & = dt\mathcal l_j f_j + ( dy-\alpha dt ) { \left(\frac{\eta_j}{\alpha } c_jf_jc_j^\dagger - f_j\right ) } \nonumber\\&\quad + \frac{dt}{2 } { \left(2c_jf_j c_j^\dagger - c_j^\dagger c_j f_j - f_jc_j^\dagger c_j\right ) } , \label{filter2}\end{aligned}\ ] ] where @xmath64 is the quantum efficiency , @xmath65 can be any positive number , and eqs .
( [ estimate2 ] ) and ( [ filter2 ] ) form a quantum filter for poissonian observations @xcite .
equation ( [ formula2 ] ) generalizes a similar classical result by snyder @xcite .
equations ( [ formula ] ) and ( [ formula2 ] ) show that continuous hypothesis testing can be done simply by comparing how the observation process is correlated with the observable estimated by each hypothesis , as schematically depicted in fig .
[ estimator - correlator ] . ) and ( [ formula2]).,scaledwidth=40.0% ] since eqs .
( [ dmz ] ) and ( [ estimate ] ) or eqs . ( [ estimate2 ] ) and ( [ filter2 ] ) have the same form as belavkin filters , one can leverage established quantum filtering techniques to update the estimates and the likelihood ratio continuously with incoming observations .
if @xmath46 has a wigner function that remains gaussian in time , the problem has an equivalent classical linear gaussian model @xcite conditioned on each hypothesis , and @xmath66 can be computed efficiently using the kalman - bucy filter , which gives the mean vector and covariance matrix of the wigner function .
the classical model also enables one to use existing formulae of chernoff bounds for classical waveform detection @xcite to bound the error probabilities .
it remains a technical challenge to compute the quantum filter for problems without a gaussian phase - space representation beyond few - level systems , but the quantum trajectory method should help cut the required computational resources by employing an ensemble of wavefunctions instead of a density matrix @xcite .
error bounds for such nonclassical problems also remain an important open problem . as an illustration of the theory ,
consider the detection of a weak classical stochastic signal , such as a gravitational wave or a magnetic field , using a quantum sensor @xcite , with @xmath1 hypothesizing the presence of the signal and @xmath67 its absence .
let @xmath68 be a vector of the state variables for the classical signal .
one way to account for the dynamics of @xmath68 is to use the hybrid density operator formalism , which includes @xmath68 as auxiliary degrees of freedom in the system @xcite .
the initial assumptive state @xmath69 becomes @xmath70 , with @xmath71 being the initial density operator for the quantum sensor and @xmath72 the initial probability density of @xmath68 .
equation ( [ dmz ] ) for @xmath73 becomes @xmath74 with @xmath75/\int dx { \operatorname{tr}}f_1 $ ] .
@xmath76 should include the lindblad generator for the quantum sensor , the coupling of @xmath68 to the quantum sensor via an interaction hamiltonian , and also the forward kolmogorov generator that models the classical dynamics of @xmath68 @xcite .
@xmath77 is an operator that depends on the actual measurement of the quantum sensor ; for cavity optomechanical force detection for example , @xmath77 is the cavity optical annihilation operator or can be approximated as the mechanical position operator if the intracavity optical dynamics can be adiabatically eliminated @xcite .
for the null hypothesis @xmath0 , the classical degrees of freedom need not be included .
@xmath78 is then @xmath71 , eq .
( [ dmz ] ) becomes @xmath79 and @xmath80 includes only the lindblad generator for the quantum sensor . in most current cases of interest in quantum
sensing , the wigner functions for @xmath81 and @xmath73 remain approximately gaussian @xcite .
kalman - bucy filters can then be used to solve eqs .
( [ detection ] ) and ( [ null ] ) for the assumptive estimates , to be correlated with the observation process according to eq .
( [ formula ] ) to produce @xmath50 , and existing formulae of chernoff bounds for classial waveform detection @xcite can be used to bound the error probabilities .
ref .
@xcite contains a simple example of such calculations .
quantum smoothing can further improve the estimation of @xmath68 @xcite in the event of a likely detection .
although smoothing is not needed here for the exact computation of @xmath50 , it may be useful for improving the approximation of @xmath50 for non - gaussian problems when the exact estimates are too expensive to compute @xcite . as a second example
, consider the test of energy quantization in a harmonic oscillator . to ensure the rigor of the test ,
imagine a classical physicist who wishes to challenge the quantum harmonic oscillator model by proposing a competing model based on classical mechanics . to devise a good classical model , he first examines quadrature measurements of a harmonic oscillator in a thermal bath . with the harmonic time dependence on the oscillator frequency removed in an interaction picture ,
the assumptive state @xmath73 for the quantum hypothesis @xmath1 obeys @xmath82 + \frac{dy}{2r}{\left(cf_1+f_1c\right ) } \nonumber\\&\quad + \frac{dt}{8q}{\left(2cf_1c - c^2f_1-f_1c^2\right ) } , \label{ho } \\ c & = q\cos\theta+p\sin\theta,\end{aligned}\ ] ] where @xmath83 is the annihilation operator , @xmath84 and @xmath85
are quadrature operators , @xmath77 is a quadrature operator with @xmath86 held fixed for each trial to eliminate any complicating measurement backaction effect , @xmath8 is the decay rate of the oscillator , and @xmath87 is a temperature - dependent parameter .
this backaction - evading measurement scheme can be implemented approximately by double - sideband optical pumping in cavity optomechanics @xcite .
an equivalent classical model for the quadrature measurements is @xmath88 where @xmath89 and @xmath90 are classical ornstein - uhlenbeck processes and @xmath91 , @xmath92 , and @xmath93 are uncorrelated classical wiener noises with @xmath94 and @xmath95 .
one can make @xmath81 diagonal and embed it with a classical distribution @xmath96 to model classical statistics ; the equation for the classical assumptive state @xmath97 is @xmath98 + \frac{dy}{r}h g_0 . \label{g0}\end{aligned}\ ] ] which is a classical duncan - mortensen - zakai ( dmz ) equation @xcite .
the assumptive estimate @xmath99 should be identical to the quantum one , as can be seen by transforming @xmath73 to a wigner function and neglecting the measurement backaction that does not affect the observations .
@xmath50 given the quadrature observations then stays at 1 , confirming that the two models are indistinguishable . in a different experiment on the same oscillator ,
the energy of the oscillator is measured instead .
let @xmath100 which can be implemented approximately by dispersive optomechanical coupling in cavity optomechanics @xcite .
@xmath73 still obeys eq .
( [ ho ] ) , but with @xmath77 now given by eq .
( [ energy_c ] ) and different @xmath42 and @xmath101 .
the measurements are again backaction - evading , as the backaction noise on the oscillator phase does not affect the energy observations .
given the prior success of the classical model , the classical physicist decides to retain eqs .
( [ x ] ) and modifies only the observation as a function of @xmath89 and @xmath90 : @xmath102 the dmz equation given by eq .
( [ g0 ] ) , assuming continuous energy , should now produce an assumptive energy estimate different from the quantum one ; it is this difference that should make the likelihood ratio increase in favor of the quantum hypothesis with more observations , if quantum mechanics is correct . previous data analysis techniques that consider only the quantum estimate @xcite fail to take into account the probability that the observations can also be explained by a continuous - energy model and are therefore insufficient to demonstrate energy quantization conclusively .
the non - gaussian nature of the problem means that bounds on the error probabilities may be difficult to compute analytically and one may have to resort to numerics , but one can also use @xmath50 as a bayesian statistic to quantify the strength of the evidence for one hypothesis against another @xcite .
discussions with j. combes , c. caves , and a. chia are gratefully acknowledged .
this material is based on work supported by the singapore national research foundation under nrf grant no .
nrf - nrff2011 - 07 . 99 e. t. jaynes , _ probability theory : the logic of science _
( cambridge university press , cambridge , england , 2003 ) ; j. m. bernardo and a. f. m. smith , _ bayesian theory _ ( wiley , new york , 2000 ) .
h. l. van trees , _ detection , estimation , and modulation theory , part i _ ( wiley , new york , 2001 ) . j. s. bell , * 38 * 447 ( 1966 ) ; s. kochen and e. p. specker , j. math .
mech .
* 17 * , 59 ( 1967 ) ; j. s. bell , _ speakable and unspeakable in quantum mechanics _ ( cambridge university press , cambridge , england , 2004 ) ; a. zeilinger , * 71 * , s288 ( 1999 ) ; a. peres , _ quantum theory : concepts and methods _ ( kluwer , new york , 2002 ) .
r. ruskov , a. n. korotkov , and a. mizel , * 96 * , 200404 ( 2006 ) ; a. palacios - laloy _ et al .
_ , nature phys .
* 6 * , 442 ( 2010 ) ; a. bednorz and w. belzig , * 105 * , 106803 ( 2010 ) .
v. b. braginsky and f. ya .
khalili , _ quantum measurement _
( cambridge university press , cambridge , england , 1992 ) ; c. m. caves _ et al . _ , * 52 * , 341 ( 1980 ) ; r. schnabel , n. mavalvala , d. e. mcclelland , and p. k. lam , nature commun . * 1 * , 121 ( 2010 ) ; the ligo scientific collaboration , nature phys .
* 7 * , 962 ( 2011 ) ; d. budker and m. romalis , nature phys . * 3 * , 227 ( 2007 ) ; c. a. muschik , h. krauter , k. hammerer , and e. s. polzik , quantum inf
. process .
* 10 * , 839 ( 2011 ) . c. w. helstrom , _ quantum detection and estimation theory _
( academic press , new york , 1976 ) .
s. lloyd , science * 321 * , 1463 ( 2008 ) ; s .- h . tan _
et al . _ , * 101 * , 253601 ( 2008 ) ; s. pirandola , * 106 * , 090504 ( 2011 ) .
a. peres , fortschr .
phys . * 48 * , 531 ( 2000 ) .
k. c. schwab and m. l. roukes , phys . today * 58 * , 36 ( 2005 ) ; t. j. kippenberg and v. j. vahala , science * 321 * , 1172 ( 2008 ) ; m. aspelmeyer , s. grblacher , k. hammerer , and n. kiesel , * 27 * , a189 ( 2010 ) ; a. d. oconnell _ et al .
_ , nature ( london ) * 464 * , 697 ( 2010 ) . j. d. thompson _
et al . _ , nature ( london ) * 452 * , 72 ( 2008 ) ; a. m. jayich _ et al . _ ,
new j. phys .
* 10 * , 095008 ( 2008 ) ; j. c. sankey , c. yang , b. m. zwickl , a. m. jayich , and j. g. e. harris , nature phys .
* 6 * , 707 ( 2010 ) .
s. haroche and j. m. raimond , _ exploring the quantum : atoms , cavities , and photons _ ( oxford university press , oxford , 2006 ) ; s. gleyzes _ et al .
_ , nature ( london ) * 446 * , 297 ( 2007 ) ; c. guerlin , _ et al . _ , nature ( london ) * 448 * , 889 ( 2007 ) ; b. r. johnson _ et al .
_ , nature phys .
* 6 * , 663 ( 2010 ) .
d. h. santamore , a. c. doherty , and m. c. cross , * 70 * , 144301 ( 2004 ) ; j. gambetta _ et al .
_ , * 77 * , 012112 ( 2008 ) ; f. helmer , m. mariantoni , e. solano , and f. marquardt , * 79 * , 052115 ( 2009 ) ; c. deng , j. m. gambetta , and a. lupacu , * 82 * , 220505(r ) ( 2010 ) .
a. a. clerk , f. marquardt , and j. g. e. harris , * 104 * , 213603 ( 2010 ) .
a. s. holevo , proc .
2nd japan - ussr symp .
prob .
theo . ,
kyoto , japan * 1 * , 22 ( 1972 ) . h. p. yuen , r. s. kennedy , and m. lax , ieee trans .
inform .
theory * it-21 * , 125 ( 1975 ) .
a. chefles , contemp .
phys . * 41 * , 401 ( 2000 ) .
s. dolinar , mit rle quart .
prog . rep .
* 111 * , 115 ( 1973 ) ; r. l. cook , p. j. martin , and j. m. geremia , nature ( london ) * 446 * , 774 ( 2007 ) ; j. gambetta , w. a. braff , a. wallraff , s. m. girvin , and r. j. schoelkopf , * 76 * , 012325 ( 2007 ) ; k. jacobs , quantum inform .
compu .
* 7 * , 127 ( 2007 ) .
b. a. chase and j. m. geremia , * 79 * , 022314 ( 2009 ) .
h. m. wiseman and g. j. milburn , _ quantum measurement and control _
( cambridge university press , cambridge , england , 2010 ) .
m. tsang , * 102 * , 250403 ( 2009 ) ; * 80 * , 033840 ( 2009 ) ; * 81 * , 013824 ( 2010 ) ; v. petersen and k. mlmer , _ ibid . _ * 74 * , 043802 ( 2006 ) ; m. tsang and c. m. caves , * 105 * , 123601 ( 2010 ) ; m. tsang , h. m. wiseman , and c. m. caves , _ ibid .
_ * 106 * , 090401 ( 2011 ) .
v. p. belavkin , radiotekh .
elektron .
( moscow ) * 25 * , 1445 ( 1980 ) ; theory probab .
appl . * 38 * , 573 ( 1993 ) ; * 39 * , 363 ( 1994 ) ; l. bouten and r. van handel , e - print arxiv : math - ph/0508006 ; l. bouten , r. van handel , and m. r. james , siam j. control optim .
* 46 * , 2199 ( 2007 ) .
t. e. duncan , inform .
control * 13 * , 62 ( 1968 ) ; t. kailath , ieee trans .
inform .
theory * it-15 * , 350 ( 1969 ) ; t. kailath and h. v. poor , _ ibid . _ * 44 * , 2230 ( 1998 ) .
see supplementary material for a detailed derivation of eqs .
( [ formula2])-([filter2 ] ) and a simple example of the application of hypothesis testing theory to quantum optomechanical force detection .
d. l. snyder , ieee trans .
inform .
theor .
* it-18 * , 91 ( 1972 ) .
h. l. van trees , _ detection , estimation , and modulation theory .
part iii _ ( wiley , new york , 2001 ) . h. j. carmichael , _ statistical methods in quantum optics 1 _ ( springer - verlag , berlin , 1999 ) ; _ statistical methods in quantum optics 2 _ ( springer - verlag , berlin , 2008 ) ; a. barchielli and m. gregoratti , _ quantum trajectories and measurements in continuous time _ ( springer - verlag , berlin , 2009 ) ; r. dum , p. zoller , and h. ritsch , * 45 * , 4879 ( 1992 ) ; j. dalibard , y. castin , and k. mlmer , * 68 * , 580 ( 1992 ) ; k. mlmer , y. castin , and j. dalibard , * 10 * , 524 ( 1993 ) ; n. gisin , * 52 * , 1657 ( 1984 ) ; helvetica phys .
acta * 62 * , 363 ( 1989 ) .
p. warszawski , h. m. wiseman , and h. mabuchi , * 65 * , 023802 ( 2002 ) ; p. warszawski and h. m. wiseman , j. opt .
b * 5 * , 1 ( 2003 ) ; * 5 * , 15 ( 2003 ) .
l. bouten , r. van handel , and a. silberfarb , j. funct .
anal . *
254 * , 3123 ( 2008 ) . j. c. crassidis and j. l. junkins , _ optimal estimation of dynamic systems _ ( chapman and hall , boca raton , 2004 ) . k. s. thorne , r. w. p. drever , c. m. caves , m. zimmermann , and v. d. sandberg , * 40 * , 667 ( 1978 ) ; j. b. hertzberg _ et al .
_ , nature phys .
* 6 * , 213 ( 2010 ) .
t. e. duncan , ph.d .
thesis , standord university , 1967 ; r. e. mortensen , ph.d .
thesis , university of california , berkeley , 1966 ; m. zakai , z. wahrsch .
verw .
geb . ,
* 11 * , 230 ( 1969 ) . |
this work has been supported by fundao de amparo pesquisa do estado de so paulo ( fapesp ) and by conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) . | we have calculated spin relaxation rates in parabolic quantum dots due to the phonon modulation of the spin - orbit interaction in presence of an external magnetic field .
both , deformation potential and piezoelectric electron - phonon coupling mechanisms are included within the pavlov - firsov spin - phonon hamiltonian .
our results have demonstrated that , in narrow gap materials , the electron - phonon deformation potential and piezoelectric coupling give comparable contributions as spin relaxation processes . for large dots ,
the deformation potential interaction becomes dominant .
this behavior is not observed in wide or intermediate gap semiconductors , where the piezoelectric coupling , in general , governs the spin relaxation processes .
we also have demonstrated that spin relaxation rates are particularly sensitive to the land @xmath0-factor . , , quantum dots , spin dephasing 73.20.dx , 63.20.kr , 71.38.+i the ability to manipulate and control processes that involve transitions between spin states is , at the moment , of extreme importance due to the recent applications in quantum computation and quantum communication .
quantum dots ( qd s ) of diverse geometries are good candidates for implementation of semiconductor quantum communication devices because the electronic , magnetic and optical properties can be controlled in the modern grown and nanofabrication techniques .
the spin dephasing time is important because it sets the length - scale on which coherent physics can be observed .
it is therefore important to understand the origin of decoherence so that ultimately it may be reduced or controlled . at the moment
remains in discussion which , between these processes , is dominant in semiconductors in low dimensional systems .
some experimental results have shown good agreement with the theoretical predictions in 2-d systems @xcite but , in general , the identification of the processes through direct comparison with the experimental results can become a formidable task .
this problem is more critical in qd s , since few experimental results exist and the theoretical discussion of the spin relaxation mechanisms is still an open subject .
khaetskii and nazarov @xcite studied two kind of processes of spin relaxation induced by phonons in gaas qd s : i ) the first mechanism is due to the dresselhaus admixture of states with an opposite spin . without the spin - orbit interaction the zeeman sublevels correspond to the orbital state with pure " spin states .
the addition of spin - orbit terms provides a small admixture of the states with opposite spin to each sublevel and , thus , enables the phonon - assistant transition between them ; ii ) the second process is related with the direct coupling between the spin and the strain field as produced by the acoustic phonons . with admixture process ( i ) , scattering rates of the order of @xmath1 s@xmath2 were obtained .
the authors also showed that the admixture process is the dominant one .
recently , woods _
et al , _ @xcite have studied spin relaxation considering the coupling between spins and phonons arising from interface motion .
this mechanism arises when interface motion due to acoustic phonons changes a parameter of the system . in this case ,
small ( @xmath3 s@xmath2 ) and strongly size dependent relaxation rates have been found by the authors . in these works , the spatial dependence of the @xmath0-factor was not considered .
also , due to the small transition energies , only the coupling by piezo - phonons were included in the rate calculation .
these approximations are not necessarily valid for inas qd s , since : i ) experimental measurements have shown that the electron @xmath0-factor depends strongly on the dot size @xcite , ii ) in inas based qd s , the zeeman transition energies can be considerably larger than for gaas qd s and , therefore , the coupling due to the deformation potential can contribute significantly to the relaxation rates .
our approach is based on the model of pavlov and firsovpavlov66,pavlov67 . in this model ,
the hamiltonian describing the transitions with spin reversal , in the scattering of electrons by phonons , can be written in a general form , @xmath4 \cdot \left ( \mathbf{p}+\frac{e}{% c}\mathbf{a}\right ) $ ] , where @xmath5 is the phonon operator , @xmath6 is the spin operator , @xmath7 is the momentum operator and @xmath8 is the vector potential for the magnetic field @xmath9 .
this interaction hamiltonian depends on spin variables and , thus , can lead to spin - flip transitions between pure spin states . in this work we have calculated spin relaxation rates in inas parabolic quantum dots , considering the phonon modulation of the spin - orbit interaction within the pavlov - firsov model .
we also study the effects of the spatial dependence of the electron @xmath0-factor and evaluate the contributions of the deformation potential and piezoelectric couplings on the spin relaxation rates .
experimental measurements and numerical calculations @xcite have indicated that in lens - shaped quasi - two dimensional self - assembled quantum dots the bound states of both electrons and holes can be understood assuming an effective parabolic potential , @xmath10 , where @xmath11 is the characteristic confinement energy and @xmath12 is the radial cylindrical coordinate . by using a one - band effective mass approximation and
considering the presence of a magnetic field @xmath13 , applied normal to plane of the dot , one can write the electron wavefunctions as @xcite @xmath14 ^{\frac{1}{2}}% \frac{\rho ^{|l|}}{a^{|l|+1}}e^{-\frac{\rho ^{2}}{2a^{2}}}e^{il\varphi } l_{n}^{|l|}\left ( { \frac{\rho ^{2}}{a^{2}}}\right ) \chi(\sigma ) .
\label{funciondeonda}\]]in the above expression @xmath15 denotes the laguerre polynomials , @xmath16 is the principal quantum number , @xmath17 is the azimuthal quantum number and @xmath18 is the spin wavefunction for spin variable @xmath19 .
the corresponding eigen - energies are
@xmath20 where @xmath21 , @xmath22 is the bohr magneton , @xmath23 is the effective length and @xmath24 .
the land @xmath0-factor and the effective mass @xmath25 are expressed in second - order @xmath26 perturbation .
@xcite @xmath27 } , \label{factorg}\]]@xmath28 } .
\label{masa}\]]here , @xmath29 is the spin - orbit splitting , @xmath30 is the energy band gap , @xmath31 is the electron energy measured from the bottom of the conduction band and @xmath32 represent the interband matrix element . for inas
the value of @xmath29 is comparable with the fundamental gap @xmath33 , thus we can expect significant variations of the electron @xmath34-factor with the size parameters . for an assisted acoustic - phonon spin - flip process
, the matrix element , @xmath35 for electron spin - flip between initial ( @xmath36 ) and final ( @xmath37 ) states with emission of a phonon of momentum @xmath38 and energy @xmath39 , can be obtained , from the pavlov - firsov spin - phonon hamiltonian @xcite as @xmath40 where @xmath41 are the spin wavefunctions quantized along the @xmath42 axis , @xmath43 is the electron envelope wavefunction , the magnetic vector potential @xmath8 is obtained in the symmetric gauge considering that the orientation of @xmath44 coincide with the @xmath42-axis .
@xmath45 , @xmath46 is the polarization vector of the longitudinal acoustic phonons , @xmath47 is the average sound velocity , @xmath48 is the mass density , @xmath49 is the system volume and @xmath50 is a coupling constant that depends on the electron - acoustic phonon coupling mechanism .
detailed expressions for the parameter @xmath50 can be found in ref .
@xcite .
the spin - flip transition rate @xmath51 is calculated from the fermi golden rule @xmath52where @xmath53 is the transition energy .
the calculations were performed at @xmath54 k and we only have considered transitions between ground state zeeman levels . the temperature dependence for one - phonon emission rate is determined from @xmath55 , where @xmath56 is the bose distribution function and @xmath57 is the rate at @xmath58 k. in the temperature regime @xmath5910 k and considering typical values of magnetic field @xmath60 , the bose function is @xmath61 and @xmath62 . for temperatures larger than few kelvin degrees , two - phonon processes should be considered as the dominant spin relaxation mechanism . these type of processes have not been considered in the present calculation . in general , we obtain that @xmath63 , @xmath64 being an integer number that depends on the electron - phonon coupling process ( @xmath64 = 7 for deformation potential and @xmath64 = 5 for piezoelectric coupling ) .
this strong dependence with the transition energy and , in consequence with the @xmath0-factor , demands that this parameter should be determined taking in account the effects of the quantum confinement .
spin splitting measurements in inas self - assembled qd s @xcite have revealed @xmath0-factors showing clear dependence on the dot size .
values in the range of 0.8 - 1.29 were reported for qd s in strong confinement regime and they differ strongly from the value of bulk inas ( @xmath65= - 14.4 ) . in order to include the spatial dependence of the @xmath0-factor in the rate calculation we have used the roth formula given in eq .
( [ factorg ] ) .
these facts are clearly illustrated in fig .
[ figure1 ] , where we show calculated inas qd spin relaxation rates , for piezoelectric and deformation potential couplings , as a function of the lateral dot size @xmath66 .
rates with @xmath67 are shown in dashed lines , and those with @xmath0 given by eq.(factorg ) , in solid lines . :
a ) piezoelectric and b ) deformation potential mechanisms .
rates calculated from eq .
( [ factorg ] ) ( for @xmath68 ) are shown in solid ( dashed ) lines . ]
our results show that the effect of confinement on the @xmath0-factor produces significant variations in the relaxation rates .
we observe that the rates obtained for @xmath68 are one order of magnitude larger than rates calculated from eq .
( [ factorg ] ) for deformation potential mechanism , @xmath13 = 1 t and @xmath66 = 5 nm .
the behavior of the rates with the qd size depends directly on the spatial overlap integral and the magnitude and on the sign of the @xmath0-factor . for negative values of @xmath0 , the rate diminishes as the dot size
is increased .
this behavior can be illustrated in the rates calculated with @xmath69 ( dashed lines in fig .
[ figure1 ] ) . the eq.(factorg ) provides positive values of @xmath0 for @xmath70 nm , in this case the rate increases as the dot size increases ( solid lines in fig .
figure1 ) . the @xmath0-factor dependence on size can be neglected in the relaxation rates , for dots with @xmath71 nm , since the energy of zeeman level becomes very small .
according with the general relation , @xmath72 , we can also observe that the rates will depend strongly on the magnetic field and can increases several order of magnitude when values of @xmath13 are swept from 0.5 t to @xmath73 t. our results demonstrate that for narrow - gap materials , the piezoelectric [ fig . [ figure1]a ) ] and deformation potential [ fig . [ figure1]b ) ] coupling present comparable contribution to the spin relaxation process . indeed , for large dots and large magnetic fields , the deformation potential interaction becomes the dominant one .
this result is not observed in gaas ( see fig . [ figure2 ] ) , where the pe coupling , in general , governs the relaxation process .
for gaas case , the confinement does not produce important modifications in the @xmath0-factor and the scattering rates shown in the fig .
[ figure2 ] reveal the negative character of the @xmath0-factor for the considered dot sizes .
for piezoelectric ( dashed lines ) and deformation potential ( solid lines ) mechanisms . ] also , the fig .
[ figure2 ] shows that the gaas rates diminish really fast as the dot size increases . for @xmath74
nm , the relaxation times can be as large as seconds and the rates become almost independent of @xmath66 .
this behavior as well as the magnitude of rates are similar to those obtained by woods @xcite when considered other relaxation mechanisms mediated by acoustic phonons . in conclusion
, we have studied the spin relaxation of electrons in inas and gaas parabolic quantum dots by considering the phonon modulation induced on the spin - orbit interaction as the relaxation process . for dots based on narrow - gap materials
, we have found that the size dependence ( spatial localization ) of the @xmath0-factor can not be neglected in the calculation .
also , the deformation potential mechanism can become dominant , especially for large @xmath13 . |
this work is supported by the iter - cn under grants nos .
2013gb104004 , 2013gb112011 and 2013gb111004 , the national science foundation of china under grant nos .
11235009 , 11205132 and 11575157 , fundamental research fund for chinese central universities and eurofusion consortium under grant agreement no .
discussion with prof .
liu chen ( zhejiang unviersity and university of california , irvine ) is kindly acknowledged .
19 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 | a new mechanism for geodesic acoustic mode ( gam ) excitation by a not fully slowed down energetic particle ( ep ) beam is analysed to explain experimental observations in large helical device .
it is shown that the positive velocity space gradient near the lower - energy end of the ep distribution function can strongly drive gam unstable .
the new features of this ep - induced gam ( egam ) are : 1 . no instability threshold in the pitch angle ; 2 .
the egam frequency can be higher than the local gam frequency ; and 3 .
the instability growth rate is much larger than that driven by a fully slowed down ep beam .
geodesic acoustic modes ( gam ) @xcite are finite frequency components of zonal structures ( zs ) , with a predominantly @xmath0 scalar potential and @xmath1 up - down asymmetric density perturbation .
gams have been intensively studied in the past two decades due to their potential roles in regulating symmetry - breaking microscopic turbulence @xcite and the associated wave - induced transport .
further to this wave - wave couplings , resonant wave - particle interactions with energetic particles ( ep ) , i.e. , the so called energetic - particle - induced gam ( egam ) , also attracted significant attention @xcite . due to their zonal " mode structures
, gam can not be excited by expansion free energy associated with pressure gradients , and velocity space anisotropy is needed for egam instability . in previous works ,
a slowing - down ep distribution with a localized pitch angle @xcite was adopted to study the egam excitation by neutral beam injection ( nbi ) , considering that ep slowing down due to collisions with electrons is much faster than pitch angle scattering due to ion collisions .
it was found that , for the fully slowed down eps , only eps with pitch angle @xmath2 are destabilizing @xcite , as later confirmed by numerical simulation @xcite . here , @xmath3 is the pitch angle , @xmath4 is the magentic moment , @xmath5 is the magnetic field at magnetic axis and @xmath6 .
berk and zhou @xcite , meanwhile , proposed that the sharp gradient in pitch angle induced by nbi prompt loss could lead to fast excitation of egam .
gam excitation by a bump - on - tail ion distribution function is addressed in refs .
, for the case where ep source is the high - energy tail in ion distribution function induced by radial frequency ( rf ) heating ; and possible interaction between egam and drift wave turbulence is discussed @xcite . a recent paper by ido et al @xcite presented experimental evidences in the large helical device ( lhd ) , where egam is observed during tangential neutral beam injection , and some peculiarities appears in the comparison with theoretical predictions .
the egam is excited before the injected nbi is fully slowed down , and the ep birth energy ( @xmath7 ) is much larger than that predicted for wave - particle resonance @xcite ; thus , earlier theories based on a fully slowed down nbi @xcite can not be directly applied here . in this work , we explain why egam is excited by a not fully slowed down ep beam , with the ep distribution function being a function of time . as we will show later , the instability drive comes from the positive velocity space gradient in the low - energy end of the ep distribution function , similar to that of a bump - on - tail distribution .
the egam excited by such an ep beam can be applied to explain the experimental observations in lhd @xcite . for the sake of clarity
, we focus only on the local dispersion relation of egam , while neglecting system nonuniformity and higher order effects , such as finite larmor radius .
modulation of ep distribution function due to excitation of egam is not taken into account either .
the dispersion relation of the @xmath8 electrostatic egam can be derived from quasi - neutrality condition , @xmath9 with @xmath10 being the nonadiabatic component of the perturbed particle distribution function @xmath11 and @xmath10 is derived from linear gyrokinetic equation@xcite @xmath12 here , the subscripts @xmath13 representing electrons , ions and eps , respectively ; and @xmath14 is the electric charge .
furthermore , @xmath15 is bessel function of the zero - index , with @xmath16 being the larmor radius , @xmath17 is the transit frequency , @xmath18 is the magnetic drift frequency ; and other notations are standard . the thermal plasma response to gam has already been derived in several earlier works , and we shall not provide the details here .
ep response can be derived by transforming into drift orbit center coordinates . since ,
in the scenario considered here , egam is excited before the eps are fully slowed down , we assume the following distribution for eps : @xmath19 here , @xmath20 is related with nbi flux and slowing - down time , @xmath21 is the heaviside step function , @xmath22 is the ep birth energy , @xmath23 is the lower end of the distribution function , @xmath24 is the slowing - down rate , and we typically have @xmath25 .
assuming small ep drift orbit normalized to egam wave length and keeping only @xmath26 transit harmonics , one then obtains the following dispersion relation from the surface averaged quasi - neutrality condition @xmath27 where @xmath28 is the gam frequency , @xmath29 and the perturbed ep density @xmath30 the dispersion relation can then be derived as @xmath31=0.\end{aligned}\ ] ] here , @xmath32 , @xmath33 , and @xmath34 and @xmath35 are the transit frequency defined at @xmath22 and @xmath36 , respectively .
note that @xmath35 is a slowly varying function of @xmath37 .
note that there are two different singularities in the dispersion relation ; i.e. , the logarithmic terms ( the first ones in the square bracket ) typical of a slowing down distribution , and the simple poles ( the second ones in the square bracket ) from the integration limits . as noted in ref . , the simple pole at @xmath38 only modulate theegam real frequency , but it does not contribute to mode excitation .
the logarithmic terms , on the other hand , are destabilizing for @xmath39 ; thus , there is a threshold , @xmath39 , for egam excitation by the fully slowed down ep beam case @xcite .
meanwhile , for the not fully slowed down distribution function considered here , the simple pole at @xmath40 is also destabilizing ; thus , there is no threshold in the pitch angle @xmath41 for egam excitation by the eps .
, width=340 ] , width=340 ] the dispersion relation can be solved numerically as a function of @xmath42 , and yield the slow temporal evolution of the excited egam due to the slowing down of the ep beam .
there are three branches ; a gam branch with @xmath43 , a lower beam branch ( lbb ) with @xmath44 ; and an upper beam branch(ubb ) , with @xmath45 . here , we take @xmath46 with the imaginary part being the landau damping rate of gam , @xmath47 , and @xmath39 such that the coefficient of the logarithmic term ( i.e. , @xmath48 ) is positive .
the real frequency and growth rate are shown , respectively , in figs .
[ fig : rf_lpa ] and [ fig : gr_lpa ] .
the horizontal axis is time in unites of @xmath49 .
we may see that , only lbb with the frequency @xmath50 can be unstable , consistent with the fact that , in the present case of experimental interest for lhd experimental observations @xcite , the ubb frequency is larger than the gam frequency @xcite .
the lbb is stable when @xmath51 ; as @xmath35 approaches @xmath28 with @xmath37 , the growth rate of lbb will increase significantly ( @xmath52 ) , and then decays quickly as @xmath36 further decreases ( @xmath53 ) .
later on ( @xmath54 ) , the growth rate decays very slowly , due to the contribution of the destabilizing logarithmic term .
so for @xmath39 , the mode can be interpreted as a double - beam plasma instability with the two singularities dominating at different times .
as @xmath35 approaches @xmath28 , the simple pole dominates ; however , as @xmath35 further decreases and becomes smaller than @xmath28 by a finite amount , the contribution from the simple pole is negligible , while the logarithmic term still contributes to destabilizing the egam @xcite .
the strong instability at @xmath55 may also provide an explanation for the fast growth of egam observed experimentally .
we note also that , the frequency of the unstable lbb , can be significantly larger than @xmath28 , as is shown in fig . [
fig : rf_lpa ] .
this may explain the higher - frequency branch of egam observed in lhd @xcite .
another evidence of that the present theory can be applied to explain the lhd experiment is that , in the discharges with electron cyclotron resonance heating ( ech ) , the dependence of the observed egam frequency on plasma temperature is weaker .
this is because that the slowing down rate is lower in the ech discharges , where collisional rate is lower due to higher electron temperature , such that the unstable mode is more a beam - branch " with the frequency determined by the transit frequency of eps @xcite .
, width=340 ] , width=340 ] the real frequency and growth rate for @xmath56 are shown in figs .
[ fig : rf_spa ] and [ fig : gr_spa ] , respectively . in this case , the logarithmic term is stabilizing @xcite ; thus , the egam discussed here is similar to beam - plasma instability , which , however , has a double pole instead of the simple pole of the present case .
we can see that , the behaviors of the three branches are similar to the case with @xmath39 .
however , when @xmath35 becomes smaller than @xmath28 by a finite amount , the growth rate of lbb decreases to zero as the contribution of the simple pole becomes vanishingly small , similar to that of a beam - plasma instability .
but it is clearly shown that , due to the contribution from the positive gradient in the lower - energy end of the distribution function , egam can still be driven unstable here , unlike the case of the fully slowed - down ep beam . in conclusion , we elucidate the mechanism for gam excitation by a not fully slowed down ep beam , which can be applied to explain the experimental observations in large helical device .
there are two kinds of singularities in the dispersion relation ; one is the logarithmic singularity typical of a slowing - down distribution function ; and the other one is simple pole from the low- and high - energy end of the distribution function .
it is found that only the lower beam branch with the frequency determined by the low - energy limit of the distribution function is unstable , consistent with the fact that the upper beam branch frequency is larger than the gam frequency as in the case of experimental interest . for @xmath39 ,
both singularities are destabilizing , with the simple pole dominating as @xmath57 ; and the contribution of the logarithmic singularity taking over as @xmath35 becomes much smaller than @xmath28 . on the other hand , for @xmath56 , only the simple pole is destabilizing , and the behavior of egam is very similar to a beam - plasma instability . in both cases , the real frequency of the unstable lower beam branch can be significantly higher than gam frequency , providing an explanation for experimental observations .
note that the real frequency of the unstable lower beam branch can be higher than local gam frequency does nt conflict with the argument we used to explain why the upper beam branch is always stable , since the simple pole at @xmath58 is always stabilizing , so one needs the unstable branch to have a frequency lower than local gam frequency for the instability of the logarithmic term ; while the lower beam branch is unstable at @xmath59 due to the contribution of the simple pole .
another novel result of this work is that , the egam drive is much stronger than that predicted in earlier theories @xcite , since the destabilizing simple pole singularity obtained in the present case is much stronger than the logarithmic singularity from a fully slowed down ep distribution function . |
we report on a measurement with the collider detector at the fermilab tevatron of the rate of pair production of top quarks in the lepton+jets channel , @xmath8 .
recent theoretical calculations predict the cross section for top pair production @xcite with an uncertainty of less than 15% .
the increase in the fermilab tevatron center - of - mass energy to 1.96 tev from 1.80 tev is expected to have enhanced the top pair production cross section by 30% .
each top quark is expected to decay into a @xmath9 boson and a @xmath10 quark , with a branching fraction of almost 100% .
a significant deviation of the observed rate of top pair production from the standard model prediction could indicate either a novel top quark production mechanism , _
e.g. _ the production and decay of a heavy resonance into @xmath11 pairs @xcite , or a novel top quark decay mechanism , _
e.g. _ a decay into supersymmetric particles @xcite , or a similar final state signature from a top - like particle @xcite .
previous measurements of the properties of the top quark @xcite are consistent with expectations from the standard model but suffer from large statistical uncertainties .
we first show that it is feasible to measure the top pair production cross section with a single kinematic event property , which may be used to discriminate between the signal from top pair production and the dominant background from @xmath9 boson production with associated jets @xcite .
this property is the total transverse energy in the event @xcite , which has been used as a discriminant by several recent top analyses @xcite .
in addition , we test the modeling of the kinematics of top pair and @xmath12jets production . a good understanding of the kinematics of these processes will be required to discover single top quark production and will benefit searches for the higgs boson and physics beyond the standard model at both the tevatron and the future large hadron collider , where techniques using kinematic discrimination have been proposed .
we then develop an artificial neural network technique in order to maximize the discriminating power available from kinematic and topological properties @xcite . throughout this paper , we quantify the gain of our neural network approach relative to the single event property of total transverse energy .
the statistical sensitivity of our neural network technique is comparable to that of methods employing secondary vertex @xmath10-tagging @xcite , and is independent of the assumptions and systematic uncertainties specific to @xmath10-tagging .
the collider detector at fermilab ( cdf ) has been substantially upgraded for the current tevatron collider run , which began in 2001 .
the major upgrades include new charged particle tracking detectors , forward calorimetry , trigger and data acquisition electronics and infrastructure as well as extended muon coverage .
a thorough description of the detector is provided elsewhere @xcite .
the essential components of the detector for this analysis are briefly described here .
the reconstruction of charged particles with high transverse momentum is essential to the electron and muon triggers that collect our data sample , the identification of electrons and muons , and the measurement of the muon momentum .
the charged particle tracking detectors are immersed in a 1.4 t magnetic field from a superconducting solenoid , which is oriented parallel to the proton beam direction @xcite .
the central outer tracker @xcite ( cot ) has eight super - layers of 310 cm long wires covering radii from 40 to 137 cm .
each super - layer consists of planes of 12 sense wires .
the super - layers alternate between having wires parallel to the cylinder axis and wires displaced by a @xmath13 stereo angle .
this provides three dimensional charged particle track reconstruction , with up to 96 position measurements with a spatial resolution of about 180 @xmath14 m in the transverse plane .
the cot transverse momentum resolution is @xmath15 @xmath16^{-1}}$ ] .
the inner tracking detector is a silicon strip detector @xcite that provides up to eight position measurements with a spatial resolution of about 15 @xmath14 m .
calorimetry is used to measure the transverse energy of electrons and jets , as well as to infer the presence of neutrinos from a significant imbalance in the observed transverse energy .
the calorimeters lie outside the solenoid and are physically divided into a central region @xcite covering pseudo - rapidity @xmath17 and an upgraded plug region @xcite covering @xmath18 .
the electromagnetic calorimeter is a lead - scintillator sandwich , which is 18 radiation lengths deep in the central region ( cem ) , with energy resolution of @xmath19 .
the hadronic calorimeter is an iron - scintillator sandwich , which is 4.5 nuclear interaction lengths deep in the central region ( cha ) , with energy resolution of @xmath20 .
the calorimeters are segmented into a projective `` tower '' geometry , where each tower subtends an area of 0.11 in @xmath21 and 15@xmath22 in azimuth in the central region .
finer position resolution for electron and photon identification is provided by proportional chambers ( ces ) , located at the approximate electromagnetic shower maximum depth in each tower .
muons are identified in drift chambers which surround the calorimeters up to @xmath23 . the central muon detector ( cmu )
@xcite consists of a set of drift chambers located outside the central hadronic calorimeters and covers @xmath24 .
an additional 60 cm thick layer of steel shields the four layers of single wire drift tubes that comprise the central muon upgrade detector ( cmp ) .
the central muon extension detector ( cmx ) consists of drift tubes , located at each end of the central detector between @xmath25 in polar angle , that extend the coverage to muons between @xmath26 .
gas cerenkov light detectors @xcite located in the @xmath27 region measure the number of inelastic @xmath28 collisions per bunch crossing and thereby the luminosity delivered to cdf by the tevatron .
the total uncertainty on the luminosity is 5.9% , where 4.4% comes from the acceptance and operation of the luminosity monitor and 4.0% from the calculation of the total @xmath28 cross section @xcite .
top quark events in the lepton+jets channel imply electron or muon of either charge . ] , @xmath29 , are characterized by a high transverse momentum lepton and substantial missing transverse energy due to the leptonic @xmath9 decay along with several hadronic jets with high transverse energy .
two jets are expected from the hadronic @xmath9 decay , two more are expected from the @xmath10 and @xmath30 quarks originating from the respective @xmath31 and @xmath32 decays . in practice , not all of these jets may be reconstructed due to kinematic requirements and limitations of the detector geometry , while other jets may arise from initial and final state hard radiation effects .
the data sample in this paper is collected by a trigger based solely on the presence of a high transverse momentum lepton . in this section ,
we discuss the trigger and lepton identification requirements , the reconstruction of the jets and the missing transverse energy , and further requirements we impose to reduce specific backgrounds .
the same criteria are applied to both data and monte carlo simulation .
this analysis uses data from @xmath28 collisions at a center - of - mass energy of @xmath33 tev collected with cdf between march 2002 and september 2003 .
all of the detector subsystems important for lepton identification and kinematic reconstruction , namely the central outer tracker , calorimeters and muon chambers , were carefully monitored over this period and any segment of data with a problem in any of these systems was excluded from consideration .
no requirement was made on the silicon detectors for this analysis .
the integrated luminosity of this data sample was measured to be 194@xmath211 pb@xmath3 @xcite .
cdf uses a three - level trigger and data acquisition system to filter interesting events from the 1.7 mhz beam crossing rate and write them to permanent storage at an average rate of 60 hz .
we describe here only the triggers important for this analysis , which select events containing an electron or muon with high transverse momentum ( @xmath34 ) .
the efficiencies of these triggers have been measured directly from the data @xcite and are listed in table [ tab : acc ] . at the first level ( l1 ) , charged particle tracks reconstructed in the cot @xmath35 projection by a hardware track processor , the extremely fast tracker ( xft ) @xcite , are required to point to a cluster of energy in the electromagnetic calorimeter or to a track segment in the muon chambers .
the l1 electron trigger requires an xft track with @xmath36 gev/@xmath37 , matched to a single trigger tower in the central electromagnetic calorimeter having transverse energy @xmath38 gev and with a ratio of hadronic to electromagnetic energy less than 0.125 .
the l1 muon trigger requires that either an xft track with @xmath39 gev/@xmath37 be matched to a muon track segment with @xmath40 gev/@xmath37 from the cmu and the cmp , or that an xft track with @xmath41 gev/@xmath37 be matched to a muon track segment with @xmath40 gev/@xmath37 from the cmx .
the second level ( l2 ) electron trigger requires the xft track matched to a cluster of energy in the central electromagnetic calorimeter with @xmath42 gev and with a ratio of hadronic to electromagnetic energy less than 0.125 .
the calorimeter cluster is formed by adding the energy in neighboring trigger towers with @xmath43 gev to the original l1 trigger tower . for this data
set , the l2 muon trigger automatically accepts events passing the l1 muon trigger . at the third level ( l3 ) , a farm of linux computers performs on - line a complete event reconstruction , including three - dimensional charged particle track reconstruction .
the l3 electron trigger requires : a track with @xmath44 gev/@xmath37 matched to a cluster of energy in three adjacent towers in pseudo - rapidity in the central electromagnetic calorimeter with @xmath45 gev ; the ratio of hadronic to electromagnetic energy less than 0.125 ; a lateral shower profile on electron identification . ] of the calorimeter cluster less than 0.4 ; and the distance between the extrapolated track position and the ces measurement in the @xmath46 view less than 10 cm .
the l3 muon trigger requires a track with @xmath47 gev/@xmath37 matched to a track segment in the muon chambers within @xmath48 cm in the @xmath35 view and , for cmu and cmp muons only , within @xmath49 cm in the @xmath46 view .
electron candidates are required to have a cot track with @xmath50 gev/@xmath37 that extrapolates to a cluster of energy with @xmath51 gev formed by three adjacent towers in pseudo - rapidity in the central electromagnetic calorimeter .
the electron energy is corrected by less than 5% for the non - uniform response across each calorimeter tower by using the ces measurement of the shower position .
the shower position is required to be away from the calorimeter tower boundaries to ensure high quality discrimination between electrons and charged hadrons .
this fiducial volume for electrons covers 84% of the solid angle in the central @xmath23 region .
the selection requirements are defined below and listed in table [ tab : elesel ] : * ratio of hadronic energy in the cluster , @xmath52 , to the electromagnetic energy in the cluster , @xmath53 .
* comparison of the lateral shower profile @xcite , the distribution of adjacent cem tower energies as a function of the seed tower energy in the calorimeter , with that expected from test beam electrons , @xmath54 .
* @xmath55 comparison of the ces shower profiles with those of test beam electrons in the @xmath46 view , @xmath56 .
* distance between the position of the extrapolated track and the ces shower profiles measured in the @xmath35 and @xmath46 views , @xmath57 and @xmath58 .
the limits on @xmath57 are asymmetric and signed by electric charge @xmath59 to allow for energy deposition from bremsstrahlung photons emitted as the electron / positron passes through the detector material
. * ratio of cluster energy to track momentum , @xmath60 . * isolation , @xmath61 , defined as the ratio between any additional transverse energy in a cone of radius @xmath62 around the cluster and the transverse energy of the cluster .
.[tab : elesel]selection requirements for electron candidates from @xmath9 boson decay . [ cols="<,<",options="header " , ]
we found the smallest expected statistical and systematic uncertainties _ a priori _ for the ann in the @xmath633 jets data sample . as a cross - check
, we repeat the analysis in the @xmath634 jet sample , where there is a higher expected signal fraction of about 42% .
we find 118 events pass the event selection criteria in our data sample with an integrated luminosity of 194 pb@xmath3 .
figures [ fig : result_ht_4j ] and [ fig : result_ann_4j ] show the distribution of data events for @xmath64 and an ann specially trained to obtain good separation in the @xmath634 jet sample .
we extract the most probable number of @xmath11 signal events : @xmath65 ( @xmath66 ) , + @xmath67 ( ann ) , where the uncertainty is statistical only and we have assumed a top mass of 175 gev/@xmath68 .
the requirement of a fourth jet with transverse energy above 15 gev reduces the @xmath11 acceptance to [email protected]% .
the measured top pair production cross section is then : @xmath69 pb ( @xmath66 ) , + @xmath70 pb ( ann ) , where the uncertainties are statistical and systematic , respectively .
we observe good agreement here between the results of the @xmath64 and ann fits . from simulated experiments
, we estimate a probability of 13% to find a difference equal to or larger than the observed difference between the results from the @xmath64 distributions in the correlated @xmath633 jets and @xmath634 jets samples . the observed 27% statistical uncertainty for the @xmath64 fit is now on the low edge of what we would expect in 68% of simulated experiments .
the observed 22% uncertainty for the ann fit is close to the median from simulated experiments .
we show the results of a fit to each of the twenty kinematic and topological properties listed in table [ tab : vardef ] for the @xmath633 jet data sample in fig .
[ fig : result_single_3j ] and the @xmath634 jets data sample in fig .
[ fig : result_single_4j ] .
some of these properties are highly correlated with each other .
we do not observe any significant difference in the results for properties used or not used by the ann .
we note that two other cdf analyses @xcite select a top sample , and measure a top cross section in the lepton+jets channel , by using a displaced secondary vertex to tag the presence of @xmath10 quarks from the @xmath71 decay .
it is of interest to see how our neural net classifies this top sample .
[ fig : ann_btag ] shows the ann output for @xmath633 jet data events with , and without , at least one @xmath10-jet identified using this secondary vertex algorithm .
the output of the network is indeed close to 1 for many of the events with at least one identified @xmath10-jet .
this provides verification that the kinematics of the @xmath10-tagged events are top - like , or , alternatively , the @xmath10-tag algorithm provides verification that the ann efficiently isolates top events using kinematics only .
we present a measurement of the top pair production cross section in @xmath72 collisions at @xmath33 tev with an integrated luminosity of 194@xmath211 pb@xmath3 .
we select events in the top lepton+jets channel by requiring one isolated lepton with @xmath73 gev , missing transverse energy @xmath74 gev , and three jets with @xmath75 gev .
this selection accepts an estimated 7.11 @xmath2 0.56% of all @xmath11 events .
we develop an artificial neural network technique , which combines the information from seven kinematic and topological properties , to discriminate between @xmath11 and background processes .
relative to the discrimination from only the total transverse energy , this artificial neural network technique reduces the expected statistical uncertainty by 30% and the estimated systematic uncertainty by 40% .
we perform a binned maximum likelihood fit to the artificial neural network output distribution observed in data , where we rely on monte carlo simulation to model the @xmath11 and @xmath9+jets processes . in a data sample of 519 events ,
we find 91@xmath216 @xmath11 events , where the uncertainty is statistical only .
we measure a top pair production cross section of @[email protected] @xmath21.5 pb , where the uncertainties are statistical and systematic , respectively .
we thank michelangelo l. mangano and claudio ferretti for their assistance with the alpgen monte carlo generator .
we thank the fermilab staff and the technical staffs of the participating institutions for their vital contributions .
this work was supported by the u.s .
department of energy and national science foundation ; the italian istituto nazionale di fisica nucleare ; the ministry of education , culture , sports , science and technology of japan ; the natural sciences and engineering research council of canada ; the national science council of the republic of china ; the swiss national science foundation ; the a.p .
sloan foundation ; the bundesministerium fr bildung und forschung , germany ; the korean science and engineering foundation and the korean research foundation ; the particle physics and astronomy research council and the royal society , uk ; the russian foundation for basic research ; the comision interministerial de ciencia y tecnologia , spain ; and in part by the european community s human potential programme under contract hprn - ct-2002 - 00292 , probe for new physics . | we present a measurement of the top pair production cross section in @xmath0 collisions at @xmath1=1.96 tev .
we collect a data sample with an integrated luminosity of 194@xmath211 pb@xmath3 with the cdf ii detector at the fermilab tevatron .
we use an artificial neural network technique to discriminate between top pair production and background processes in a sample of 519 lepton+jets events , which have one isolated energetic charged lepton , large missing transverse energy and at least three energetic jets .
we measure the top pair production cross section to be @xmath4@xmath5 pb , where the first uncertainty is statistical and the second is systematic . * measurement of the cross section for @xmath6 production in @xmath0 collisions using the kinematics of lepton+jets events * = cmti8 # 1@xmath7 d. acosta , j. adelman , t. affolder,9 t. akimoto , m.g .
albrow , d. ambrose , s. amerio , d. amidei , a. anastassov , k. anikeev , a. annovi , j. antos,1 m. aoki , g. apollinari , t. arisawa , j - f .
arguin , a. artikov , w. ashmanskas , a. attal,7 f. azfar , p. azzi - bacchetta , n. bacchetta , h. bachacou , w. badgett , a. barbaro - galtieri , g.j .
barker , v.e .
barnes , b.a .
barnett , s. baroiant,6 g. bauer , f. bedeschi , s. behari , s. belforte , g. bellettini , j. bellinger , a. belloni , e. ben - haim , d. benjamin , a. beretvas , t. berry , a. bhatti , m. binkley , d. bisello , m. bishai , r.e .
blair,2 c. blocker,5 k. bloom , b. blumenfeld , a. bocci , a. bodek , g. bolla , a. bolshov , d. bortoletto , j. boudreau , s. bourov , b. brau,9 c. bromberg , e. brubaker , j. budagov , h.s .
budd , k. burkett , g. busetto , p. bussey , k.l .
byrum,2 s. cabrera , m. campanelli , m. campbell , f. canelli,7 a. canepa , m. casarsa , d. carlsmith , r. carosi , s. carron , m. cavalli - sforza,3 a. castro,4 p. catastini , d. cauz , a. cerri , l. cerrito , j. chapman , y.c . chen,1 m. chertok,6 g. chiarelli , g. chlachidze , f. chlebana , i. cho , k. cho , d. chokheli , j.p .
chou , s. chuang , k. chung , w - h .
chung , y.s .
chung , m. cijliak , c.i .
ciobanu , m.a .
ciocci , a.g .
clark , d. clark,5 m. coca , a. connolly , m. convery , j. conway,6 b. cooper , k. copic , m. cordelli , g. cortiana , j. cranshaw , j. cuevas , a. cruz , r. culbertson , c. currat , d. cyr , d. dagenhart,5 s. da ronco , s. dauria , p. de barbaro , s. de cecco , a. deisher , g. de lentdecker , m. dellorso , s. demers , l. demortier , m. deninno,4 d. de pedis , p.f .
derwent , c. dionisi , j.r .
dittmann , p. dituro , c. drr , a. dominguez , s. donati , m. donega , j. donini , m. donofrio , t. dorigo , k. ebina , j. efron , j. ehlers , r. erbacher,6 m. erdmann , d. errede , s. errede , r. eusebi , h - c .
fang , s. farrington , i. fedorko , w.t .
fedorko , r.g .
feild , m. feindt , j.p .
fernandez , r.d .
field , g. flanagan , l.r .
flores - castillo , a. foland , s. forrester,6 g.w .
foster , m. franklin , j.c .
freeman , y. fujii , i. furic , a. gajjar , m. gallinaro , j. galyardt , m. garcia - sciveres , a.f .
garfinkel , c. gay , h. gerberich , d.w .
gerdes , e. gerchtein , s. giagu , p. giannetti , a. gibson , k. gibson , c. ginsburg , k. giolo , m. giordani , m. giunta , g. giurgiu , v. glagolev , d. glenzinski , m. gold , n. goldschmidt , d. goldstein,7 j. goldstein , g. gomez , g. gomez - ceballos , m. goncharov , o. gonzlez , i. gorelov , a.t .
goshaw , y. gotra , k. goulianos , a. gresele , m. griffiths , c. grosso - pilcher , u. grundler , j. guimaraes da costa , c. haber , k. hahn , s.r .
hahn , e. halkiadakis , a. hamilton , b - y .
han , r. handler , f. happacher , k. hara , m. hare , r.f .
harr , r.m .
harris , f. hartmann , k. hatakeyama , j. hauser,7 c. hays , h. hayward , b. heinemann , j. heinrich , m. hennecke , m. herndon , c. hill,9 d. hirschbuehl , a. hocker , k.d .
hoffman , a. holloway , s. hou,1 m.a .
houlden , b.t .
huffman , y. huang , r.e .
hughes , j. huston , k. ikado , j. incandela,9 g. introzzi , m. iori , y. ishizawa , c. issever,9 a. ivanov,6 y. iwata , b. iyutin , e. james , d. jang , b. jayatilaka , d. jeans , h. jensen , e.j .
jeon , m. jones , k.k .
joo , s.y .
jun , t. junk , t. kamon , j. kang , m. karagoz unel , p.e .
karchin , y. kato , y. kemp , r. kephart , u. kerzel , v. khotilovich , b. kilminster , d.h .
kim , h.s .
kim , j.e .
kim , m.j .
kim , m.s .
kim , s.b .
kim , s.h .
kim , y.k .
kim , m. kirby , l. kirsch,5 s. klimenko , m. klute , b. knuteson , b.r .
ko , h. kobayashi , d.j .
kong , k. kondo , j. konigsberg , k. kordas , a. korn , a. korytov , a.v .
kotwal , a. kovalev , j. kraus , i. kravchenko , a. kreymer , j. kroll , m. kruse , v. krutelyov , s.e .
kuhlmann,2 s. kwang , a.t .
laasanen , s. lai , s. lami , s. lammel , m. lancaster , r. lander,6 k. lannon , a. lath , g. latino , i. lazzizzera , c. lecci , t. lecompte,2 j. lee , j. lee , s.w .
lee , r. lefvre,3 n. leonardo , s. leone , s. levy , j.d .
lewis , k. li , c. lin , c.s .
lin , m. lindgren , e. lipeles , t.m .
liss , a. lister , d.o .
litvintsev , t. liu , y. liu , n.s .
lockyer , a. loginov , m. loreti , p. loverre , r - s .
lu,1 d. lucchesi , p. lujan , p. lukens , g. lungu , l. lyons , j. lys , r. lysak,1 e. lytken , d. macqueen , r. madrak , k. maeshima , p. maksimovic , g. manca , f. margaroli,4 r. marginean , c. marino , a. martin , m. martin , v. martin , m. martnez,3 t. maruyama , h. matsunaga , m. mattson , p. mazzanti,4 k.s .
mcfarland , d. mcgivern , p.m. mcintyre , p. mcnamara , r. mcnulty , a. mehta , s. menzemer , a. menzione , p. merkel , c. mesropian , a. messina , t. miao , n. miladinovic,5 j. miles , l. miller , r. miller , j.s .
miller , c. mills,9 r. miquel , s. miscetti , g. mitselmakher , a. miyamoto , n. moggi,4 b. mohr,7 r. moore , m. morello , p.a .
movilla fernandez , j. muelmenstaedt , a. mukherjee , m. mulhearn , t. muller , r. mumford , a. munar , p. murat , j. nachtman , s. nahn , i. nakano , a. napier , r. napora , d. naumov , v. necula , j. nielsen , t. nelson , c. neu , m.s .
neubauer,8 t. nigmanov , l. nodulman,2 o. norniella,3 t. ogawa , s.h .
oh , y.d .
oh , t. ohsugi , t. okusawa , r. oldeman , r. orava , w. orejudos , k. osterberg , c. pagliarone , e. palencia , r. paoletti , v. papadimitriou , a.a .
paramonov , s. pashapour , j. patrick , g. pauletta , m. paulini , c. paus , d. pellett,6 a. penzo , t.j .
phillips , g. piacentino , j. piedra , k.t .
pitts , c. plager,7 l. pondrom , g. pope , x. portell,3 o. poukhov , n. pounder , f. prakoshyn , a. pronko , j. proudfoot,2 f. ptohos , g. punzi , j. rademacker , m.a .
rahaman , a. rakitine , s. rappoccio , f. ratnikov , h. ray , b. reisert , v. rekovic , p. renton , m. rescigno , f. rimondi,4 k. rinnert , l. ristori , w.j .
robertson , a. robson , t. rodrigo , s. rolli , r. roser , r. rossin , c. rott , j. russ , v. rusu , a. ruiz , d. ryan , h. saarikko , s. sabik , a. safonov,6 r. st .
denis , w.k .
sakumoto , g. salamanna , d. saltzberg,7 c. sanchez,3 l. santi , s. sarkar , k. sato , p. savard , a. savoy - navarro , p. schlabach , e.e .
schmidt , m.p .
schmidt , m. schmitt , t. schwarz , l. scodellaro , a.l .
scott,9 a. scribano , f. scuri , a. sedov , s. seidel , y. seiya , a. semenov , f. semeria,4 l. sexton - kennedy , i. sfiligoi , m.d .
shapiro , t. shears , p.f .
shepard , d. sherman , m. shimojima , m. shochet , y. shon , i. shreyber , a. sidoti , a. sill , p. sinervo , a. sisakyan , j. sjolin , a. skiba , a.j .
slaughter , k. sliwa , d. smirnov , j.r .
smith,6 f.d .
snider , r. snihur , m. soderberg , a. soha,6 s.v .
somalwar , j. spalding , m. spezziga , f. spinella , p. squillacioti , h. stadie , m. stanitzki , b. stelzer , o. stelzer - chilton , d. stentz , j. strologas , d. stuart,9 j. s. suh , a. sukhanov , k. sumorok , h. sun , t. suzuki , a. taffard , r. tafirout , h. takano , r. takashima , y. takeuchi , k. takikawa , m. tanaka,2 r. tanaka , n. tanimoto , m. tecchio , p.k .
teng,1 k. terashi , r.j .
tesarek , s. tether , j. thom , a.s .
thompson , e. thomson , p. tipton , v. tiwari , s. tkaczyk , d. toback , k. tollefson , t. tomura , d. tonelli , m. tnnesmann , s. torre , d. torretta , w. trischuk , r. tsuchiya , s. tsuno , d. tsybychev , n. turini , f. ukegawa , t. unverhau , s. uozumi , d. usynin , l. vacavant , a. vaiciulis , a. varganov , s. vejcik iii , g. velev , v. veszpremi , g. veramendi , t. vickey , r. vidal , i. vila , r. vilar , i. vollrath , i. volobouev , m. von der mey,7 p. wagner , r.g .
wagner,2 r.l .
wagner , w. wagner , r. wallny,7 t. walter , z. wan , m.j .
wang,1 s.m .
wang , a. warburton , b. ward , s. waschke , d. waters , t. watts , m. weber , w.c .
wester iii , b. whitehouse , d. whiteson , a.b .
wicklund,2 e. wicklund , h.h .
williams , p. wilson , b.l .
winer , p. wittich , s. wolbers , c. wolfe , m. wolter , m. worcester,7 s. worm , t. wright , x. wu , f. wrthwein,8 a. wyatt , a. yagil , t. yamashita , k. yamamoto , j. yamaoka , c. yang , u.k .
yang , w. yao , g.p .
yeh , j. yoh , k. yorita , t. yoshida , i. yu , s. yu , j.c .
yun , l. zanello , a. zanetti , i. zaw , f. zetti , j. zhou , and s. zucchelli,4 .026 in ( cdf collaboration ) .026 in 1 institute of physics , academia sinica , taipei , taiwan 11529 , republic of china + 2 argonne national laboratory , argonne , illinois 60439 + 3 institut de fisica daltes energies , universitat autonoma de barcelona , e-08193 , bellaterra ( barcelona ) , spain + 4 istituto nazionale di fisica nucleare , university of bologna , i-40127 bologna , italy + 5 brandeis university , waltham , massachusetts 02254 + 6 university of california , davis , davis , california 95616 + 7 university of california , los angeles , los angeles , california 90024 + 8 university of california , san diego , la jolla , california 92093 + 9 university of california , santa barbara , santa barbara , california 93106 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + |
in cosmological models at present time the universe is assumed to have various ingredients of exotic nature like the dark energy candidates , among which we find the phantom scalar field @xcite .
the reason is that supernovae ia data allow an equation of state of the dark energy component with @xmath2 , where @xmath3 and @xmath4 are the pressure and the energy density of the fluid , which is a condition that a phantom scalar field satisfies at cosmic scales .
this is a reason to start up an exploration of the consequences such type of scalar field might have at local scales , for instance on black holes . on the other hand , the behavior of the area of the horizon when the black holes is accreting matter is a very important property of black hole physics , because the accretion of such exotic material may impose important restrictions on the mass of black nowadays black hole candidates . in order to study this process in the full non - linear regime
we start up with a model coupling the phantom scalar field and gravity .
the model assumes the lagrangian density of the phantom scalar field is given by @xmath5 where @xmath6 is the ricci scalar of the space - time , @xmath7 is the space - time metric , @xmath8 is a scalar field and @xmath9 is the potential of the field .
the property defining a phantom scalar field is that the relative sign between the ricci scalar and the kinetic term are the same .
when the action constructed with such a lagrangian density is varied with respect to the metric , the arising einstein s equations are related to a stress energy - tensor that violates the null energy condition , that is @xmath10 , where @xmath11 is a null vector .
the immediate implication of this violation is the violation of the weak energy condition too , which in turn implies that observers following time - like trajectories might measure negative energy densities .
although this property is at odds with the nowadays physics observed in laboratories , there are cosmological observations indicating the presence of such kind of matter@xcite . in this paper
we explore the implications of this at astrophysical scales .
the fact that the scalar field violates the null energy condition motivates the study of possible unusual implications in astrophysical scenarios , because the area increasing theorem does not apply in this case . for instance , using exact solutions corresponding to stationary accretion of a test phantom fluid , it was found that the mass of black holes decreases in a phantom energy dominated universe approaching the big rip @xcite .
it has also been studied recently the behavior of a black hole apparent horizon ( ah ) in a frw background , and the conditions under which a naked singularity can be formed due to the coincidence of the ah of the black hole and the cosmic horizon @xcite ; in such case the authors consider the effects of the back reaction of the scalar field on the space - time metric and indicate that cosmic censorship does not only forbid the existence of naked singularities but also the existence of a phantom field . assuming that the null energy condition is satisfied the area of the horizon only increase .
however , if this condition is violated , the area of the event horizon decreases and the black hole shrinks as shown recently in the full non - linear regime using numerical relativity in @xcite . the present paper is a detailed follow up of previous one , in which we also explore the effects of the scalar field potential on the accretion rates and final state of the black hole .
two important items are presented in this paper : i ) the relation @xmath12 is not fulfilled at local scales by the scalar field ( at least near to a black hole ) although the null energy condition is not satisfied and ii ) the accretion of such scalar field reduces the area of a black hole at similar rates for different types of potentials driving the scalar field .
this paper is organized as follows . in section
[ sec : cauchy ] we describe the 3 + 1 decomposition of the space - time , the evolution system of equations driving the evolution of the geometry and the construction of the initial data . in section [ sec : results ] we present the results obtained . finally in section
[ sec : conclusions ] we draw some conclusions and comments .
we formulate einstein s field equations coupled to the scalar field in such a way that these can be integrated numerically . in this model
, we will use spherical symmetry and use geometrized units for which the speed of light and newton s constant are equal to one . in order to the study the dynamics of a spherically symmetric space - time , we write the general metric for the coordinate system @xmath13 in the following form @xmath14 , \label{eq : metric}\end{aligned}\ ] ] where @xmath15 acts as a conformal factor relating this metric to a space - like flat metric , @xmath16 is the only non - zero component of the shift vector and @xmath17 is the lapse function as in @xcite .
we solve the evolution einstein s field equations using the generalized bssn evolution formulation of the 3 + 1 decomposition of general relativity described in @xcite as opposed to previous successful analyzes refereed to the accretion of scalar field using eddington - finkelstein like coordinates under the adm formulation @xcite . for the construction of the space - time
we carry out a cauchy - type evolution of initial data based on the 3 + 1 decomposition of the space - time . within such decomposition
there are various formulations of the evolution equations with different hyperbolic properties . among such formulations
the most popular nowadays is the so called bssn formulation @xcite which helped at solving the problem of the binary black hole collision system recently @xcite using the punctures technique that allowed the adequate treatment of the black hole singularity @xcite .
the bssn formulation assumes that the conformal metric has determinant equal to one , but in spherical coordinates the flat metric the determinant is different to one . this issue is addressed relaxing this condition over the determinant , obtaining the generalized bssn equations ( gbssn ) @xcite , which is the formulation we use in our simulations . the gbssn system of equations in a spherically symmetric space - time reduce to a set of six equations for the independent dynamical variables @xmath15 , @xmath18 , @xmath19 , the non - zero trace - free part of the extrinsic curvature @xmath20 , the trace of the extrinsic curvature @xmath21 and the contracted christoffel non - zero symbol ( conformal connection function ) @xmath22 .
the explicit expressions in the presence of matter are : @xmath23 where primes denote derivatives with respect to @xmath24 and the gauge parameter @xmath25 is such that for @xmath26 the coordinates are eulerian , whereas for @xmath27 the coordinates are lagrangian .
we have introduced the matter sources @xmath4 , @xmath28 , @xmath29 and @xmath30 which represent the energy density , the momentum current density , the non - zero trace free part of the stress tensor and the trace of the stress tensor respectively .
these quantities can be computed after projecting the stress - energy tensor onto the unit normal vector @xmath31 to the space - like hyper - surfaces , @xmath32 the constraints of the system are : the hamiltonian constraint @xmath33 , the momentum constraint @xmath34 and the constraint arising from the definition of the conformal connection functions @xmath35 . in spherical symmetry
they are : @xmath36 in order to specify the gauge , we evolve the lapse according to the standard 1+@xmath37 slicing condition @xmath38 or the related condition obtained by dropping the advection term @xmath39 . for the shift vector we implemented the recipe for the @xmath40-driver condition @xmath41 which helps to avoid the slice stretching near the horizon , which is known to kill the numerical evolution . in order to avoid instabilities near the origin ( the puncture )
, we implement a sort of excision without excision @xcite using a factor function on the source of the evolution equations of the form @xmath42 from the coordinate origin out to @xmath43 an appropriate fraction of the size of the apparent horizon radius such that @xmath43 is smaller that the apparent horizon of the final black hole .
even though this function violates the constrains , the violations do not propagate out of the black hole horizon as shown by the convergence tests . in all our simulations we use eulerian coordinates ,
that is , we set @xmath27 .
the evolution of the scalar field is driven by the klein - gordon equation @xmath44 -\frac{dv}{d\phi}=0 .
\label{eq : wave}\ ] ] in order to reduce this equation to a first order system as the geometric counterpart , we define two new variables @xmath45 and @xmath46 , and ( [ eq : wave ] ) reduces to the system @xmath47 \nonumber\\ & + & \partial_r \left [ \left(\frac{\alpha^2 \chi}{g_{rr } } - \beta^r\right)\psi + \beta^2 \pi \right ] + \alpha^2 \frac{dv}{d\phi } \nonumber\\ \partial_t \psi & = & \partial_r \pi .
\label{eq : sf_evolution}\end{aligned}\ ] ] which together with the evolution equations for the geometry and the completes the system of equations to be solved .
in order to start up an evolution including the matter terms it is necessary to solve the constraints a the initial time slice , and afterwards bianchi identities guarantee that such constraints are satisfied .
we assume that the initial slice is time - symmetric , which implies that the shift vector @xmath48 and its time derivative are zero initially .
in addition , time symmetry imply that all the components of the extrinsic curvature are zero initially . in this way , the momentum constraint is satisfied identically at initial time . on the other hand
, we use the pre - collapsed condition for the lapse @xmath49 . in order to provide a local nature to the scalar field
we assume that @xmath8 has a gaussian profile or is a train of gaussians one next to the other , which implies initial data for the first order variable @xmath50 , and the time symmetry provides the condition @xmath51 . with this information about the matter source
we solve the hamiltonian constraint .
now , in order to achieve a smooth coordinate system we use the initial ansatz that the space - time metric has a form similar to that of a schwarzschild black hole in isotropic coordinates .
thus we assume that the metric at initial time has the form @xmath52 where @xmath53 is the mass of the apparent horizon and @xmath54 is the function to be determined through the solution of the hamiltonian constraint .
the hamiltonian constraint is thus reduced to the equation @xmath55 which we solve using a fourth order runge - kutta integrator . in the last equation ( [ eq : u ] ) , @xmath9 is the self - interaction potential of the phantom scalar field .
we finally rescaled the coordinate @xmath24 in such way that @xmath56 when @xmath57 .
these initial data are used to start up a cauchy type evolution using the gbssn evolution equations ( [ eq : gbbsn_evolution ] ) , the gauge conditions ( [ eq : lapse ] - [ eq : shift ] ) and the scalar field equation ( [ eq : sf_evolution ] ) . the numerical method used to approximate the constraint and evolution equations is a fourth order finite differences approximation . we only perform the evolution on a finite domain with artificial boundaries at a finite value of @xmath24 where we implement radiative - type boundary conditions . the integration in time
uses a method of lines with a fourth order accurate runge - kutta integrator . throughout the evolution
, we monitor the hamiltonian , momentum and @xmath58 constraints , so that we check that they converge to zero as resolution is increased with fourth order .
_ apparent horizons_. in order to track the radius , area and mass of the apparent horizon during the evolution in terms of the @xmath59 variables , we search for the marginally trapped surfaces ( mts ) through the condition @xmath60 where we take @xmath61 to be the outward pointing unit vector normal to the horizon , @xmath62 are the components of the extrinsic curvature and @xmath21 its trace of a space - like hypersurface on which one calculates the mtss .
the apparent horizon is the outermost mts . in our coordinates equation ( [ eq : mts ] ) for
the metric ( [ eq : metric ] ) reads @xmath63 in order to track the evolution of the apparent horizon we calculate @xmath64 at every time step and locate the outermost zero of it at the coordinate radius @xmath65 .
then calculate the area of the corresponding 2-sphere @xmath66 and its mass @xmath67 , where @xmath68 is the areal radius evaluated at @xmath65 .
_ misner - sharp mass function _ this quantity is a space - dependent measure of the mass of the space - time .
we use this function in order to compare the mass of the apparent horizon and the mass of the space - time , so that we can monitor the consistency of our simulations . the misner - sharp mass function in our case
is written as @xmath69 , \end{aligned}\ ] ] where @xmath70 .
the misner - sharp mass reduces to the arnowitt - deser - misner ( adm ) at space - like infinity , @xmath71 so we can estimate the mass of the space - time at a finite radius at any time during the evolution . on the other hand
, the misner - sharp mass function reduces to the bondi - sachs mass at future null infinity .
_ event horizon_. aside of the apparent horizon which is a gauge dependent 2-surface , we track a bundle of radial outgoing null rays in order to approximately locate the event horizon during the evolution as the 3-surface from which outgoing null rays diverge when launched toward the future .
for this we solve the geodesic equation for radial null rays on the fly during the evolution , and fine - tune their location initially so that both , outgoing null rays escaping toward future null infinity and outgoing null rays that reach the singularity converge to the same surface for as much time as possible during the evolution , after which they spread .
such surface is the approximate location of the event horizon .
in fig . [ fig : id ] we show a particular case of an initial scalar field profile .
we can see that the solution of the hamiltonian constraint implies already at initial time a negative energy density as expected from the violation of the null energy condition of this type of field .
this also opens another possibility to be explored , that is , once we know that the scalar field contributes with a negative energy density it is also possible to construct initial configurations whose adm mass is negative , that is , the contribution of the scalar field energy density to the space - time is bigger in absolute value than that due to the black hole s horizon . on the other hand , we also show in fig . [
fig : id ] the relation @xmath72 where @xmath73 , which clearly indicates that at local scales , the lagrangian of the phantom scalar field does not behave at all as phantom matter since @xmath74 , and in fact runs from cosmological constant @xmath75 to stiff matter @xmath76 .
in our study we set up two classes of initial data corresponding to positive adm mass and negative adm mass for all the potentials studied . in order to reduce the parameter space of our study , for each potential we fine - tune the initial scalar field profile so that the adm mass has the same value for all the potentials explored .
the potential we used are : a ) quadratic : @xmath77 , b ) quartic : @xmath78 and c ) exponential : @xmath79 , where @xmath80 and @xmath81 are negative constant parameters .
we also explored the potentials @xmath82 and @xmath83 , however the behavior with these potentials is very similar to the obtained with the potentials mentioned before . in fig .
[ fig : massadm ] we show the accretion of the scalar field for all the potentials , in particular , we show the mass of the apparent horizon and the misner - sharp masses .
the finding is that despite of the potential used , the final mass of the apparent horizon is nearly the same in all cases , that is , the potential has no effect on the final state of the resulting black hole . for both classes of configurations , positive and negative adm mass , we found that the apparent horizon mass decreases and converges asymptotically to the misner - sharp mass . as a particular example , we show in fig .
[ fig : massadm ] the behavior of the apparent horizon mass , the misner - sharp mass and the even horizon location for the exponential potential case . in order to make sure that the decrease of the black hole is real ,
we fine - tuned a bundle of outgoing null rays for the positive and negative adm mass cases , and estimated the location of the event horizon , since it is gauge invariant .
we show the results in fig .
[ fig : massadm ] where we present the location of the event horizon for the exponential potential case .
the behavior shown in fig .
[ fig : massadm ] is generic and applies to all the cases studied , that is , the event horizon -not only the apparent horizon- shrinks and is located approximately at the apparent horizon surfaces when the system has stabilized . finally , in order to validate our numerical results we show in fig .
[ fig : pconver ] the convergence of the @xmath84 norm of the constraints to zero as the resolution is increased with fourth order for two representative cases with the exponential potential . in fig .
[ fig : masscom ] we summarize the results related to the accretion mass rate and final mass of the apparent horizon for all various potential used . in each case
we fine - tuned different scalar field profiles for the different potentials in order to use only one value of the adm mass for the positive case and one value for the negative case .
we present the full non - linear spherical accretion of a phantom scalar field with various potentials into a black hole .
this accretion process is very efficient at reducing the black hole horizon area , which is a potentially important process that may impose restrictions on the existing black hole masses , or for instance primordial black holes . in our simulations we were able to reduce the mass of the initial black hole up to 50@xmath85 .
attempts to reduce the area even further presents difficulties with our approach due to the steep gradients in the equations near the origin , and other techniques might be implemented in order to sort this problem out .
we show that the final mass of the black hole s horizon is nearly independent of the potential used .
this results is physically interesting because the process of area reduction will apply independently of the cosmologically motivated potential used .
we show that even though the lagrangian ( [ eq : lagrangian ] ) of the scalar field corresponds to a phantom field , at local scales the radial equation of state is always @xmath86 , and it depends on space and time , with values ranging from a cosmological constant equation of state @xmath75 up to a stiff mater @xmath87 .
we want to stress the importance of the equation of state of the scalar field given by the lagrangian ( [ eq : lagrangian ] ) , and point out that perhaps the big rip scenario would strongly depend on the homogeneity and isotropy of the scalar field .
we have shown how much the equation of state can depend on space and time for the model of a scalar field , at least in the strong gravitational field regime .
the astrophysical possibility of this accretion process strongly depends on the chance of having a phantom scalar field located near a black hole with a rather sharp profile .
this work was supported in part by grants cic - umsnh 4.9 and 4.23 , promep umich - ptc-210 , umich - ca-22 and umich - ca-22 network from sep mexico , and conacyt grant numbers 79601 and 106466 .
the runs were carried our in the ifm cluster .
m. shibata , t. nakamura , evolution of three dimensional gravitational waves : harmonic slicing case .
d * 52 * 5428 , 1995 . t. w. baumgarte , s. l. shapiro , on the numerical integration of einstein s field equations .
d * 59 * 024007,1999 . | we present the accretion of a phantom scalar field into a black hole for various scalar field potentials in the full non - linear regime .
our results are based on the use of numerical methods and show that for all the cases studied the black hole s apparent horizon mass decreases .
we explore a particular subset of the parameter space and from our results we conclude that this is a very efficient black hole shrinking process because the time scales of the area reduction of the horizon are short . we show that the radial equation of state of the scalar field depends strongly on the space and time , with the condition @xmath0 , as opposed to a phantom fluid at cosmic scales that allows @xmath1 . |
the study of very young massive stars is of particular importance for star formation since it is not yet firmly established by what process , disk accretion or stellar merging , such stars form ( e.g. , * ? ? ?
* ; * ? ? ?
previous authors have demonstrated that massive stars can be formed via accretion , leading to the formation of large ( @xmath3 au ) and massive ( @xmath4 m@xmath0 ) circumstellar disks . however , most of these stars are located more than 2 kpc away , which makes it challenging to confirm _ unambiguously _ the presence of accretion disks around single objects with radio interferometry due to the lack of spatial resolution .
we focus on afgl 490 one of the best - studied young stars that are in a transition stage to herbig be stars .
this very young star is located nearby ( @xmath5 kpc , * ? ? ?
* ) , has a mass of 810 m@xmath0 , a bolometric luminosity of @xmath6 l@xmath0 ( b23 star , * ? ? ?
* ) , is still embedded in its parent molecular cloud ( a@xmath7 mag , * ? ? ?
* ) , and drives a high - velocity outflow @xcite .
interferometric measurements by e.g. ( * ? ? ?
* hereafter paper i ) as well as ir observations of @xcite and @xcite have revealed that the star is located inside an ionized @xmath8 au cavity cleared up by a fast stellar wind and surrounded by a @xmath9 au disk - like structure , which is enshrouded in a 22000 au @xmath10 6000 au envelope .
all these structures are further embedded in an even more extended envelope . in this letter , we present subarcsecond - resolution pdbi observations of afgl 490 that prove the existence of a rotating disk around afgl 490 and determine its orientation , size , and mass by using comprehensive physical , chemical , and line radiative transfer modeling .
we observed afgl 490 in c@xmath1o ( 21 ) at 224.714389 ghz using the pdbi in its b configuration in december 2003 ( baselines 61.5m330.5 m ) .
the phase reference center of our measurements was @xmath11 = 03@xmath12 27@xmath13 38@xmath1455 , @xmath15 = @xmath16 46@xmath17 59@xmath188 , and the source velocity was set to @xmath19 kms@xmath20 .
for the line measurements , we used one correlator unit with a total bandwidth of 20 mhz and 512 channels ( velocity resolution of 0.052 kms@xmath20 ) .
two other correlator units with a bandwidth of 320 mhz each were utilized to measure the continuum . the bandpass and phase calibration were performed on the objects 3c454.3 and 2145 + 067 .
the additional calibrations of the phase and amplitude were obtained by observing the objects 0355 + 508 and 0224 + 671 every 20 minutes . for the final phase calibration and data reduction
the grenoble software environment gildas was applied .
the maps of @xmath21 square pixels with @xmath22 pixel size were produced by the fourier transform of the calibrated visibilities using natural weighting .
the synthesized beam size is @xmath23 ( @xmath24 au at 1 kpc ) with a positional angle of 73@xmath25 . for reduction of the continuum
, we excluded the line emission from the data . finally , the continuum was subtracted from the original data in the @xmath26-plane , as it was detected only at the source position .
within the @xmath27 primary pdbi beam at 1.3 mm , afgl 490 was detected point source with a total flux of 1.41 jy ( peak value is 0.68 jybeam@xmath20 ) .
the continuum is dominated by the thermal dust emission because at 1.3 mm the free - free radiation contributes less than 5% to the total flux ( @xmath28 mjy , see * ? ? ?
the integrated c@xmath1o ( 21 ) line intensity map overlaid with the 1.3 mm continuum is shown in fig .
the continuum is barely resolved at the @xmath29 level within the @xmath30 au area around the star , and has a beam - like circular shape .
in contrast , the line intensity map reveals a @xmath31 au arc of emission that is centered on the continuum peak and has a pa of @xmath2 @xmath32 .
furthermore , the c@xmath1o emission above the 50% intensity level appears as a @xmath33 700 au @xmath10 2000 au bar - like structure with the main axis shifted by 400 au to the northeast from the center .
the red- and blue - shifted parts of the integrated c@xmath1o ( 21 ) line intensity overlaid with an @xmath34-band speckle image @xcite and the 1.3 mm continuum is presented in fig .
both lobes have their maximum 700 au away from the 1.3 mm continuum point source , and the line connecting them is shifted 260 au to the northeast ( pa of the axis perpendicular to the connecting line is @xmath35 ) . due to the simultaneous measurement of both line and continuum in the same bandpass ,
the positional offset is certainly real .
remarkably , the position of the red- and blue - shifted c@xmath1o ( 21 ) peaks perfectly coincides with the position of the cs ( 21 ) peaks reported in paper i. the pa of the c@xmath1o emission is smaller than the @xmath36 positional angle of the @xmath37 au bar - like structure ( see paper i ) , but the emission structure is orthogonal to the large - scale co outflow @xcite and coincides with the pa obtained from the infrared polarized map and the 2 cm radio emission @xcite .
note that the c@xmath1o emission pattern ( fig .
[ co]b ) is characteristic of outflows or rotating disk - like configurations . to distinguish between these two possibilities , we analyze the corresponding position - velocity diagram below .
we estimate the mass of the afgl 490 `` inner disk '' from the 1.3 mm continuum emission , using eq . 2 from @xcite .
we assume the standard gas - to - dust mass ratio of 100 , a mean disk temperature of @xmath38 k , and a dust mass absorption coefficient @xmath39 at 1.3 mm between 1.99 and 5.86 @xmath40g@xmath20 ( grains without icy mantles , gas densities 10@xmath41 and 10@xmath42 @xmath43 , respectively , see * ? ? ?
the resulting h@xmath44 gas mass is between 0.8 and 2.3 m@xmath0 and , thus , the source - averaged h@xmath44 column density is @xmath45 @xmath46 .
the gas mass can be also derived from the c@xmath1o data . due to missing flux of @xmath230% in the vicinity of afgl 490 , even in the reverse map side of the red- and the blue - shifted line emission , the total integrated flux value is smaller than the sum of individual contributions from the red- and blue - shifted emission .
the sidelobes of the synthesized beam are not contributing to the low integrated fluxes due to their low intensities ( @xmath4720% ) .
thus , we focus on the sum of the red- and the blue - shifted emission . assuming optically thin c@xmath1o line emission , the gas mass can be derived by @xmath48 & = & 1.58\,10^{-10 } \
\ \frac { \tau \ ( t_{\rm ex}+0.9 ) \ { \rm exp}(16.18/t_{\rm ex } ) } { ( 1-{\rm exp}(-\tau ) ) } \\
\nonumber & & \times \ { x({\rm c^{17}o } ) } \ ( { \rm d[kpc]})2 \int_{line}\!\!\!\!s_{\nu } \delta{\rm v \ [ jy \ km\,s^{-1 } ] } \ , \end{aligned}\ ] ] where @xmath49 is the optical depth of the c@xmath1o ( 21 ) line , @xmath50 k is the excitation temperature , @xmath51 is the h@xmath44 abundance relative to the amount of c@xmath1o molecules ( h@xmath44/co=@xmath52 , @xmath53o/@xmath1o@xmath54 ) , and @xmath55 kpc is the distance to the source .
the masses of the red- and blue - shifted gas are 1.0 and 0.6 m@xmath0 , respectively , giving a total value of @xmath56 m@xmath0 which matches the estimate obtained above . the gas velocity distribution across the red- and blue - shifted c@xmath1o emission ( solid line in fig . [ co]b )
is presented in the position - velocity ( pv ) diagram in fig .
the pv map shows a number of single gas clumps , indicating an inhomogeneous disk structure .
this distribution is not caused by optical depth effects as demonstrated by radiative transfer calculations ( see sect .
3.3 ) . to fit the pv diagram
, we adopt a simple model of a keplerian disk with a mass that linearly increases with radius @xcite and a stellar mass of @xmath57 m@xmath0 .
we assume a disk radius of 1600 au based on the spatial extent of the c@xmath1o emission and a disk mass between 0.5 and 2 m@xmath0 . with this model , we fit the inner clumps ( @xmath58 ) of the the pv map and obtain the best - fit inclination angle of @xmath59 ( fig .
[ vco ] , solid curve ) .
the fit to the most intense clumps in the pv map at @xmath60 is not as good . it gives @xmath61 and super - keplerian gas velocities in the inner disk .
in contrast to massive disks around active galactic nuclei that show a modified keplerian law with an exponent of @xmath2 @xmath62 ( e.g. , * ? ? ?
* ) , the corresponding curve for afgl 490 ( fig . [ vco ] , dotted curve ) would only fit to the outer disk part .
however , it does not fit the inner clumps .
furthermore , the four most intense clumps in the pv map around @xmath63 kms@xmath20 can be best fitted by a rotating gas ring with a radius of 700 au and similar @xmath64 ( dashed line ) .
such a small inclination angle is in contrast to the apparent c@xmath1o morphology implying an inclination angle larger than 60@xmath25 for any circularly - symmetric configuration , e.g. a disk ( see ellipse in fig .
however rotation curves for any disk or ring model with larger inclination angles , @xmath65 , can only be well fitted assuming a low - mass central star , @xmath66 2 m@xmath0 which is definitely not the case for afgl 490 . thus , we propose that there is a massive , clumpy disk around afgl 490 with nearly face - on orientation ( @xmath67 ) and keplerian rotation .
the pdbi c@xmath1o ( 21 ) spectral map of the afgl 490 disk is presented in fig .
[ c17o_fit ] .
we fit these spectra using the `` step - by - step '' modeling approach ( * ? ? ?
* for details see semenov et al .
2006 , in preparation ) .
briefly , we apply the 2d flared - disk model of @xcite to simulate the disk physical structure , a gas - grain chemical network to calculate time - dependent abundances , and a 2d line radiative transfer code of @xcite to synthesize the c@xmath1o spectra .
we assume that the disk has the power - law surface density @xmath68 and the velocity profile @xmath69 .
the dust grains have mrn - like size distribution and the gas - to - dust mass ratio is 100 .
moreover , we assume an age of @xmath20.1 myr due to the lack of reliable estimates . @xcite reported the dynamical outflow time of 1.8 10@xmath70 yr which is an lower limit ( see * ? ? ?
the best fit to the observed c@xmath1o ( 21 ) lines is obtained with the following disk parameters : ( 1 ) inclination and positional angles of @xmath71 , ( 2 ) inner and outer radii of @xmath72 au and @xmath73 au , respectively , ( 3 ) a surface density gradient @xmath74 , and ( 4 ) a keplerian - like velocity profile , @xmath75 .
the best - fit disk mass ranges between @xmath76 and 1 m@xmath0 .
note that the modeled line profiles are indeed optically thin , @xmath77 .
all these values are in a good agreement with the observational results , and the values derived from the pv diagram .
why do we see such a clumpy disk structure ?
this object seems to be relatively young and could still be in a perturbed state that remains from an earlier , non - steady accretion phase .
the viscous dissipation timescale for the afgl 490 disk is estimated to be @xmath2 1 myr @xcite , which far exceeds its age .
this hypothesis is further supported by the fact that the @xmath78 years old , large - scale co outflow consists of single moving gas clumps , and thus accretion is indeed non - steady @xcite .
another explanation could be that the c@xmath1o ( 21 ) emission traces the densest parts of the spiral arms that are easily excited in massive disks by gravitational instabilities .
indeed , the morphology of the c@xmath1o emission resembles the density structure of a self - gravitating disk with double arms ( upper corner in fig . [ co]a ) calculated by @xcite .
moreover , the numerical simulations by @xcite imply that such disks should also develop a dual structure composed of _ an inner thin keplerian disk _ fed by a thicker self - gravitating torus with nearly uniform rotation .
finally , a clumpy disk structure could be a result of a recent encounter with a nearby star or due to the gravitational interaction with a wide low - mass companion .
the complex multiple outflow systems seen in the close vicinity of afgl 490 strongly support this idea ( see fig .
4b in paper i ) . in this letter , we present clear evidence that @xmath79 m@xmath0 stars can be surrounded by keplerian disks in their earliest evolutionally phase ( see also * ? ? ? * ) .
in contrast to the previously reported huge disks around such stars with radii of @xmath3 au ( e.g. , chini et al . 2004 , but see sako et al .
2005 ) the afgl 490 disk is much smaller , @xmath80 au .
using advanced theoretical modeling , we constrain basic disk parameters : ( 1 ) the inclination and positional angles are @xmath81 , ( 2 ) the surface density profile is @xmath82 , ( 3 ) the mass is @xmath2 1.5 m@xmath0 ( @xmath83 uncertainty ) , and ( 4 ) the disk rotation is close to kepler s law . | we observed the embedded , young 810 m@xmath0 star afgl 490 at subarcsecond resolution with the plateau de bure interferometer in the c@xmath1o ( 21 ) transition and found convincing evidence that afgl 490 is surrounded by a rotating disk . using two - dimensional modeling of the physical and chemical disk structure coupled to line radiative transfer
, we constrain its basic parameters . we obtain a relatively high disk mass of 1 m@xmath0 and a radius of @xmath2 1500 au .
a plausible explanation for the apparent asymmetry of the disk morphology is given . |
mathematical models for scientific and engineering systems often involve with some uncertainties .
we may roughly classify such uncertainties into two kinds .
the first kind of uncertainties may be called _
model uncertainty_. they involve with physical processes that are less known , not yet well understood , not well - observed or measured , and thus difficult to be represented in the mathematical models .
the second kind of uncertainties may be called _
simulation uncertainty_. this arises in numerical simulations of multiscale systems that display a wide range of spatial and temporal scales , with no clear scale separation . due to the limitations of computer power , at present and for the conceivable future
, not all scales of variability can be explicitly simulated or resolved .
although these unresolved scales may be very small or very fast , their long time impact on the resolved simulation may be delicate ( i.e. , may be negligible or may have significant effects , or in other words , uncertain ) .
thus , to take the effects of unresolved scales on the resolved scales into account , representations or parameterizations of these effects are desirable . these uncertainties are sometimes also called _ unresolved scales _ , as they are not represented or not resolved in modeling or simulation .
model uncertainties have been considered in , for example , @xcite and references therein .
works relevant for parameterizing unresolved scales include @xcite , among others .
in this paper we consider an issue of approximating model uncertainty or simulation uncertainty ( unresolved scales ) by stochastic processes , and then devise a stochastic scheme for such approximations .
we first recall some basic facts about fractional brownian motion ( fbm ) in [ fbm ] .
then we discuss model uncertainty and simulation uncertainty in [ modelerror ] and [ sles ] , respectively . finally , we present an example in [ example ] demonstrating our result .
this example involves approximating subgrid scales via correlated noises , in the context of large eddy simulations of a partial differential equation .
we discuss a model of colored noise in terms of fractional brownian motion ( fbm ) , including a special case which is white noise in terms of usual brownian motion .
the fractional brownian motion @xmath0 , indexed by a so called hurst parameter @xmath1 , is a generalization of the more well - known process of the usual brownian motion @xmath2 .
it is a centered gaussian process with stationary increments .
however , the increments of the fractional brownian motion are not independent , except in the usual brownian motion case ( @xmath3 ) . for more details , see @xcite
. + + definition of fractional brownian motion : for @xmath4 , a gaussian process @xmath0 , or @xmath5 , is a fractional brownian motion if it starts at zero @xmath6 , has mean zero @xmath7 = 0 $ ] , and has covariance @xmath8 = \frac12(|t|^{2h } + |s|^{2h } - |t - s|^{2h})$ ] for all t and s. the standard brownian motion is a fractional brownian motion with hurst parameter @xmath3 . + + some properties of fractional brownian motion : a fractional brownian motion @xmath9 has the following properties : + ( i ) it has stationary increments ; + ( ii ) when @xmath10 , it has independent increments ; + ( iii ) when @xmath11 , it is neither markovian , nor a semimartingale .
+ we use the weierstrass - mandelbrot function to approximate the fractional brownian motion .
the basic idea is to simulate fractional brownian motion by randomizing a representation due to weierstrass . given the hurst parameter @xmath12 with @xmath13 , we define the function @xmath14 to approximate the fractional brownian motion : @xmath15 where @xmath16 is a constant , @xmath17 s are normally distributed random variables with mean @xmath18 and standard deviation @xmath19 , and the @xmath20 s are uniformly distributed random variables in the interval @xmath21 .
the underlying theoretical foundation for this approximation can be found in @xcite .
figures 1 and 2 show a sample path of the usual brownian motion ( i.e. , @xmath3 ) , and fractional brownian motion with hurst parameter @xmath22 , respectively . _ , width=384,height=288 ] , with @xmath23_,width=384,height=288 ]
we consider a spatially extended system modeled by a partial differential equation ( pde ) : @xmath24 where @xmath25 is a linear ( unbounded ) differential operator , and @xmath26 is a nonlinear function of @xmath27 with @xmath28 and @xmath29 , and satisfies a local lipschitz condition .
in fact , @xmath26 may also depend on the gradient of @xmath30 .
if this ( deterministic ) model is accurate , i.e. , its prediction on the field @xmath30 matches with the observational data @xmath31 on a certain period of time @xmath32 $ ] , then there is no need for a stochastic approach .
however , when the prediction @xmath30 deviates from the observational data @xmath31 , we then need to modify the model . in this case
, the observational data @xmath31 may be thought to satisfy a modified model : @xmath33 where the model uncertainty @xmath34 is usually a fluctuating ( i.e. , random ) process , as the observational data @xmath31 is so ( i.e. , has various samples or realizations ) . the model discrepancy or model uncertainty
@xmath34 may have various causes , such as missing physical mechanisms ( not represented in the deterministic model ) .
sometimes , the model uncertainty @xmath34 is smaller in magnitude than other terms in the model and thus is often ignored in the deterministic modeling .
however , being small and being fluctuating may not necessarily imply that its impact on the overall system evolution to be small @xcite . to take this impact into account
, we would like to model or approximate @xmath34 by a stochastic process .
we first calculate the model uncertainty @xmath34 via observational data @xmath31 . by discretizing and using data samples for @xmath31 ,
we obtain ( discretized ) samples for @xmath35 .
the time correlation may then be calculated using the samples of @xmath35 .
if the time correlation scale is significantly shorter than the time scale for the field @xmath30 , we may ignore the time correlation and thus approximate @xmath35 by the following stochastic process containing a ( uncorrelated ) white noise , for example : @xmath36 where @xmath37 is the mean of @xmath35 ( computed from data ) , @xmath38 is the usual brownian motion ( reviewed in [ example ] below ) and the deterministic noise intensity @xmath39 may depend on space . here
@xmath39 may be computed via stochastic calculus , especially the ito isometry , as follows .
@xmath40 ^ 2 & = & \sigma^2 \;{{\mathbb e}}[\int_0^t \tilde{u } db_t]^2,\\ { { \mathbb e}}[\int_0^t ( f - { { \mathbb e}}f ) dt]^2 & = & \sigma^2 \;{{\mathbb e}}\int_0^t \tilde{u}^2(x , t ) dt,\\\end{aligned}\ ] ] thus we obtain @xmath41 ^ 2}{{{\mathbb e}}\int_0^t \tilde{u}^2(x , t ) dt } } \ ; .\end{aligned}\ ] ] with this approximation , we obtain the following stochastic partial differential equation ( spde ) as a modified model for the original deterministic model : @xmath42 @xmath43 in general , the model uncertainty @xmath35 may be better approximated by correlated noise via fractional brownian motion . since the procedure is similar , we will demonstrate this in the next section when we discuss simulation uncertainty .
this section deals with simulation uncertainty , i.e. , stochastically parameterizing the effects of the unresolved scales on the resolved scales .
we consider this issue in the context of large eddy simulations ( les ) of a nonlinear partial differential equation with memory .
in large eddy simulations of fluid or geophysical fluid flows @xcite , the unresolved scales appear as the so - called subgrid scales ( sgs ) .
the sgs term appears to be highly fluctuating ( random " ) ; see the figure 1 in @xcite .
partially motivated by this , stochastic parameterizations of subgrid scales have been investigated in fluid , geophysical and climate simulations , based on physical or intuitive or empirical arguments .
another , perhaps more important , motivation for applying stochastic parameterizations of subgrid scales is to induce the desired backward energy flux ( stochastic backscatter " ) in fluid simulations @xcite .
we present one stochastic parameterization scheme of the subgrid scale term in the large eddy simulation of a nonlinear partial differential equation with an extra memory term , which is in fact a nonlinear integro - partial differential equation .
the approximation scheme is based on stochastic calculus involving with a fractional brownian motion , and the parameter to be calculated is a spatial function , which is derived using ito stochastic calculus .
@xmath44 where @xmath25 is a linear differential operator , and @xmath35 is a nonlinear function of @xmath27 with @xmath28and @xmath29 , and satisfies a local lipschitz condition .
we investigate stochastic parameterizations of unresolved scales in the context of large eddy simulations of the above system .
the idea of large eddy simulation is to split the flow into a local , spatial mean ( or average ) and a fluctuation about the that mean .
the mean @xmath45 is defined by filtering or mollification ( convolution with an approximate identity ) .
the goal is to predict the mean accurately .
this is widely believed possible based on the idea that since fluctuations have random character , their average effects on the mean notion can successfully be medelled .
to filter the solution , we pick a filter .
many different ones are commonly used . to fix the ideas , in this paper
, we use gaussian filter as in @xcite , @xmath46 where @xmath47 is the filter size and the filter is such that : ( i ) @xmath48 is infinitely differentiable in space and , ( ii ) @xmath49 as @xmath50 in @xmath51 . here and hereafter @xmath52 or the over bar @xmath45 denotes convolution .
the mean @xmath45 is a weighted average of @xmath30 about the point @xmath53 . as @xmath54 ,
the points near @xmath53 are weighted more and more heavily , so @xmath49 as @xmath54 in @xmath51 .
using the fact that convolution commutes with differentiatian , we get the space - filtered system : @xmath55 or @xmath56 where the subgrid scale term @xmath57 . since generally @xmath58 , the usual parameterization or closure problem of the large eddy simulation has arisen . due to
inaccurate ( uncertain ) initial conditions or boundary conditions , @xmath59 is a correlated fluctuating process @xcite , depending on samples @xmath60 in a suitable sample space @xmath61 .
we thus would like to approximate the subgrid scale term @xmath62 by a stochastic process with a correlated ( i.e. , colored ) noise component , for example : @xmath63 where @xmath64 is a colored noise ( generalized time derivative of a fractional brownian motion ; reviewed in
[ example ] below ) , and @xmath65 is the mean component of the subgrid scale term @xmath66 .
moreover , the noise intensity @xmath67 is a non - negative deterministic function to be determined from fluctuating sgs data @xmath68 .
the subgrid scale term @xmath66 may be inferred from observational data ( see @xcite for relevant information for subgrid scales in navier - stokes equations ) , or from fine mesh simulations . note that @xmath39 is to be calculated or estimated from the fluctuating sgs data @xmath68 , either from observation or from fine mesh simulations .
so this is an inverse problem . as in usual inverse problems
@xcite , the stochastic parameterizations for the sgs term @xmath68 is not unique .
what we proposed above is merely an example .
this offers an opportunity for trying various stochastic parameterization schemes , much as one uses various smoother functions ( e.g. , polynomials or fourier series ) to approximate less regular functions or data in deterministic approximation theory . to estimate the unknown parameter ( function ) @xmath67 ,
we start with the following relation : @xmath69 taking time integral over a computational interval @xmath32 $ ] on both sides , we obtain @xmath70dt = \int_0^t \sigma(x ) d{b}^h_t = \sigma(x ) b^h_t.\end{aligned}\ ] ] therefore , taking mean - square on both sides , @xmath71dt)^2 & = & \sigma^2(x ) t^{2h}.\end{aligned}\ ] ] thus an estimator for @xmath67 is @xmath72dt)^2 } \;\ ; , \label{ax2}\end{aligned}\ ] ] which can be computed numerically . by the stochastic parameterization ( [ noise2 ] ) on the sgs term @xmath68 , with @xmath73 determined from ( [ force2 ] ) and @xmath74 from ( [ ax2 ] ) , the les model ( [ les2 ] ) becomes a stochastic partial differential equation ( spde ) for the large eddy solution @xmath75 : @xmath76 with the appropriately filtered boundary condition and filtered initial condition .
we present a specific example of stochastic modeling of simulation uncertainty of subgrid scales , in the context of large eddy simulations . we consider the following nonlinear partial differential equation with a memory term ( time - integral term ) @xcite : @xmath77 under appropriate initial condition @xmath78 and boundary conditions @xmath79 with @xmath80 constants , on a bounded domain @xmath81 . here
@xmath82 is a positive constant .
this model arises in mathematical modeling in ecology @xcite , heat conduction in certain materials @xcite and materials science @xcite . the time - integral term here represents a memory effect depending on the past history of the system state , and this memory effect decays polynomially fast in time .
the large eddy solution @xmath83 is the true solution @xmath30 looked through a filter : i.e. , through convolution with a spatial filter @xmath84 , with spatial scale ( or filter size or cut - off size ) @xmath47 : @xmath85 in this paper , we use a gaussian filter as in @xcite , @xmath86 . we can write @xmath92 with @xmath45 the large eddy term and @xmath93 the fluctuating term .
note that @xmath94 .
so the sgs term @xmath90 involves nonlinear interactions of fluctuations @xmath93 and the large eddy flows .
thus @xmath90 may be regarded as a function of @xmath45 and @xmath93 : @xmath95 .
the leads to a possibility of approximating @xmath90 by a suitable stochastic process defined on a probability space @xmath96 , with @xmath97 , the sample space , @xmath98field @xmath99 and probability measure @xmath100 .
this means that we treat @xmath68 data as random data as in @xcite , which take different realizations , e.g. , due to fluctuating observations or due to numerical simulation with initial and boundary conditions with small fluctuations . in fluid or geophysical fluid simulations
, the sgs term may be highly fluctuating and time - correlated @xcite , and this term may be inferred from observational data @xcite , or from fine mesh simulations .
this further suggests for parameterizing the subgrid scale term @xmath90 as a time - correlated or colored noisy term .
the increments of fractional brownian motion are correlated in time and hence its generalized time derivative @xmath101 is used as a model for colored noise . in the special case @xmath3 , we have the white noise @xmath102 .
thus we parameterize the subgrid scale term @xmath103 , which is time - correlated , by colored noise @xmath101 as follows : @xmath104 where @xmath105 is the mean component of the subgrid scale term @xmath103 .
moreover , the noise intensity @xmath67 is a non - negative deterministic function to be determined from fluctuating sgs data @xmath68 .
the subgrid scale term @xmath90 may be inferred from observational data @xcite , or from fine mesh simulations as we do here .
we represent the mean component @xmath106 in terms of the large eddy solution @xmath45 .
the specific form for @xmath73 depends on the nature of the mean of @xmath68 .
here we take @xmath107 , where coefficients @xmath108 s are determined via data fitting by minimizing @xmath109 ^ 2dx dt$ ] .
moreover , we take @xmath110 as a scalar fractional brownian motion .
note that @xmath39 is to be calculated or estimated from the fluctuating sgs data @xmath68 , either from observation or ( in this paper ) from fine mesh simulations ; see detailed discussions in @xcite .
so this is an inverse problem . as in usual inverse problems
@xcite , the stochastic parameterizations for the sgs term @xmath68 is not unique .
this offers an opportunity for trying various stochastic parameterization schemes , much as one uses various smoother functions ( e.g. , polynomials or fourier series ) to approximate less regular functions or data in deterministic approximation theory . to estimate the unknown parameter ( function ) @xmath67 , we start with - to get the following relation : @xmath111 taking time integral over a computational interval @xmath32 $ ] on both sides
, we obtain @xmath112dt = \int_0^t \sigma(x ) d{b}^h_t = \sigma(x ) b^h_t.\end{aligned}\ ] ] therefore , taking mean - square on both sides , @xmath113dt)^2 & = & \sigma^2(x ) t^{2h}.\end{aligned}\ ] ] thus an estimator for @xmath67 is @xmath114dt)^2 } \;\ ; , \label{ax}\end{aligned}\ ] ] which can be computed numerically . by the stochastic parameterization
( [ noise ] ) on the sgs term @xmath68 , with @xmath73 determined from ( [ force ] ) and @xmath74 from ( [ ax ] ) , the les model ( [ les ] ) becomes a stochastic partial differential equation ( spde ) for the large eddy solution @xmath75 : @xmath115 with boundary conditions @xmath116 and filtered initial condition @xmath117 fine mesh simulations of the original system with memory ( [ heat ] ) are conducted to generate benchmark solutions or solution realizations , with initial conditions slightly perturbed ; see fig .
these fine mesh solutions @xmath30 are used to generate the sgs term @xmath68 defined in ( [ ef ] ) at each time and space step . the filter size used in calculating @xmath68
is taken as @xmath119 .
the mean @xmath73 is calculated from ( [ force ] ) via cubic polynomial data fitting ( as discussed in the last section ) , and parameter function @xmath67 is calculated as in ( [ ax ] ) . the stochastic les model ( [ new ] )
is solved by the same numerical code but on a coarser mesh .
note that a four times coarser mesh simulation with no stochastic parameterization for the original system ( [ heat ] ) does not generate satisfactory results ; see fig .
4 . the stochastic les model ( [ new ] )
is then solved in the mesh four times coarser than the fine mesh used to solve the original equation ( [ heat ] ) .
the stochastic parameterization leads to better resolution of the solution as shown in fig .
5 . as in @xcite
, it can be shown that when two stochastic parameterization terms are close in mean - square norm on finite time intervals , the solutions are also close in the same norm .
l. arnold , hasselmann s program visited : the analysis of stochasticity in deterministic climate models . in j .-
von storch and p. imkeller , editors , _ stochastic climate models_. pages 141158 , boston , 2001 .
birkhuser .
t. n. palmer , g. j. shutts , r. hagedorn , f. j. doblas - reyes , t. jung and m. leutbecher . representing model uncertainty in weather and claimte prediction .
earth planet .
sci . _ * 33 * ( 2005 ) , 163 - 193 .
c. penland and p. sura , sensitivity of an ocean model to details " of stochastic forcing . in _ proc .
ecmwf workshop on represenation of subscale processes using stochastic - dynamic models_. reading , england , 6 - 8 june 2005 . | model uncertainties or simulation uncertainties occur in mathematical modeling of multiscale complex systems , since some mechanisms or scales are not represented ( i.e. , unresolved " ) due to lack in our understanding of these mechanisms or limitations in computational power .
the impact of these unresolved scales on the resolved scales needs to be parameterized or taken into account .
a stochastic scheme is devised to take the effects of unresolved scales into account , in the context of solving nonlinear partial differential equations .
an example is presented to demonstrate this strategy .
* key words : * stochastic partial differential equations ( spdes ) ; stochastic modeling ; impact of unresolved scales on resolved scales ; model error ; large eddy simulation ( les ) ; fractional brownian motion * mathematics subject classifications ( 2000 ) * : 60h30 , 60h35 , 65c30 , 65n35 |
the electronic transport properties of disordered materials depend essentially on the physical symmetry of the system under examination
. the well known standard symmetry classes orthogonal , unitary , and symplectic are appropriate for describing situations that occur in the presence or absence of time reversal symmetry , and broken spin rotational symmetry , respectively .
recently , an additional chiral symmetry has been attracted considerable interest for disordered two - dimensional systems @xcite .
this special symmetry may appear if the underlying lattice is bi - partite @xcite , at which eigenfunctions can occupy only one of the two sub - lattices . in two - dimensional lattice models with diagonal disorder ,
the chiral symmetry is not observed so that for orthogonal and unitary symmetry all electronic states are localized .
however , for non - diagonal disorder situations , like the motion of quantum particles subject to random magnetic fluxes as investigated in the present contribution , the eigenvalues occur in pairs @xmath1 , a characteristic feature of the chiral symmetry @xcite . appertaining to the chiral symmetry
is a critical point in the center of the tight - binding band with a diverging localization length at energy @xmath2 @xcite and a scale independent critical conductance .
we consider a two - dimensional tight - binding lattice model where the perpendicular random magnetic field is introduced via complex hopping terms .
these are chosen such that the magnetic flux per plaquette is given by the sum of the random peierls phases along the two bonds in the @xmath3-direction @xmath4 .
the corresponding hamiltonian with nearest neighbor hopping is @xmath5 where the width @xmath6 ( @xmath7-direction ) and the length @xmath8 ( @xmath3-direction ) of the sample is given in units of the lattice constant @xmath9 .
the @xmath10 and @xmath11 create or annihilate a fermi particle at the site @xmath12 , respectively .
the random fluxes are distributed uniformly , @xmath13 , with probability density @xmath14 , and the randomness is maximal for @xmath15 used in the following .
the average magnetic flux through the system is zero .
we calculate the dimensionless two - terminal conductance @xcite for systems of length @xmath8 via @xmath16 here , @xmath17 is the transmission matrix with the @xmath18 parameterizing its eigenvalues .
@xmath19 is the number of open channels .
dirichlet boundary conditions are imposed in the transversal direction . on the square root of the aspect ratio @xmath20 ( length to width ) .
] for odd ( @xmath21 ) and even ( @xmath22 ) sample widths . ] .
the solid lines are given by the function ( [ fit ] ) .
note the divergence of @xmath23 for @xmath24 when @xmath6 is odd . ]
fig . [ rmf - fig12 ] confirms the theoretical prediction @xcite that for quasi one - dimensional systems ( q1d ) , @xmath25 decreases much slower for @xmath6 being odd than for the same system with @xmath6 even .
this is easy to understand from eq .
( [ gg ] ) and from the mean values of the parameters @xmath18 . in systems with chiral symmetry ,
the @xmath18 assume both positive and negative values .
when ordered in absolute values , @xmath26 , it follows from symmetry reasons that @xmath27 for @xmath6 odd , but @xmath28 when @xmath6 is even .
this is confirmed in fig .
[ 65 - 350a ] which shows the probability distribution @xmath29 both for @xmath6 odd and even . for @xmath6
even , @xmath29 converges to the gaussian distribution with non - zero mean when the length @xmath8 of the system increases , but @xmath29 remains half - gaussian for @xmath6 odd .
the different form of the distribution of @xmath30 influences the transmission through very long q1d systems and thus the conductance .
[ rmf - plng-1 ] shows the distribution of the logarithm of the conductance @xmath23 . in very long q1d systems ,
the conductance is determined by the first channel , @xmath31 , so that @xmath32 we find @xmath33 where @xmath34 $ ] , and @xmath35 . for @xmath6
even , @xmath36 is gaussian with a non - zero mean value .
consequently , @xmath23 is gaussian , too .
the singularity of @xmath37 in eq .
( [ fit ] ) is eliminated by the exponential decrease of @xmath36 . however , for @xmath6 odd , @xmath36 possesses a maximum at @xmath38 so that @xmath23 is singular for @xmath24 .
since this singularity survives for any length of the system , there is a non - zero probability to obtain a sample with conductance @xmath39 of order of unity .
this explains the absence of exponential localization in these systems , we find @xmath40 for @xmath6 odd instead as shown in fig .
[ rmf - fig12 ] .
single parameter scaling theory requires that the conductance is a function of only one parameter in the vicinity of the critical point , @xmath41.\ ] ] here @xmath42 is a correlation length which diverges as @xmath43 . in ref .
@xcite we have analyzed the scaling of the smallest lyapunov exponent of q1d systems with _
odd _ width and found the critical exponent @xmath44 .
as also shown in ref .
@xcite , no scaling analysis of the lyapunov exponent is possible for @xmath6 even because of strange finite size effects .
therefore , we verify in this paper the one parameter scaling for 2d square samples @xmath45 with even @xmath6 using the mean conductance ( fig .
[ g - scal ] ) .
first , we subtract from the conductance the finite size correction , given by the irrelevant scaling term @xmath46 , @xmath47 where @xmath48 and @xmath49 .
then , the data scale to the universal curve , @xmath50 $ ] with a critical exponent @xmath0 , consistent with our recent result for the divergence of the localization length in samples having odd width @xcite . from the conductance , all data with @xmath51 scale to a universal curve . ]
the two - terminal conductance of a random flux model has been investigated numerically .
it shows critical behavior at the band center due to the chiral unitary symmetry of the system .
the critical exponent obtained from a finite size scaling analysis agrees with the one that governs the divergence of the localization length .
these results support the notion of universality at the chiral critical point . | the two - terminal conductance of a random flux model defined on a square lattice is investigated numerically at the band center using a transfer matrix method . due to the chiral symmetry
, there exists a critical point where the ensemble averaged mean conductance is scale independent .
we also study the conductance distribution function which depends on the boundary conditions and on the number of lattice sites being even or odd . we derive a critical exponent @xmath0 for square samples of even width using one - parameter scaling of the conductance .
this result could not be obtained previously from the divergence of the localization length in quasi - one - dimensional systems due to pronounced finite - size effects . and critical conductance , random flux , chiral symmetry , critical exponent , localization length 73.23.-b , 71.30.+h , 72.10.-d |
( * a * ) @xmath64 is displayed with a color scale given by the side bar .
the abscissa is time @xmath65 , and the ordinate is space @xmath17 .
( * b * ) the generated stationary pattern @xmath66 and @xmath67 . the abscissa is the concentration of @xmath66 or @xmath67 , and the ordinate is space @xmath17.,width=566 ] ( black solid ) and @xmath4 ( black dotted ) as well as the limit cycle orbit ( red ) are plotted for the model ( 2 ) with @xmath68 .
( * b * ) state space plot of the nullcline for x ( @xmath69 ) are plotted with the change in space @xmath17 , following the change in the diffusion term @xmath70 , where the cross - points of the nullclines give @xmath71 .
the nullclines are shifted horizontally with the spatial gradient .
, width=566 ] of the reaction diffusion ( [ eq : example1 ] ) and the predicted pattern of @xmath72 obtained from the spatial map ( [ eq : spatialmap ] ) .
the predicted pattern ( given by @xmath73 ) agrees well with the stationary pattern ( lines ) .
three patterns with boundary values of @xmath74 ( purple ) , @xmath75 ( green ) , and @xmath76 ( yellow ) are considered .
, width=566 ] of the emergent pattern and the period of the limit cycle orbit @xmath49 is shown by changing parameter @xmath24 .
results from different @xmath5 values are also plotted with different symbols .
the relationship from the brusselator eq .
( [ eq : brusselator ] ) is also plotted from the obtained data by varying @xmath77.,width=566 ]
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * sf1 * ( * a * ) the real part of the jacobi matrix eigenvalues around the uniform solution for each wavenumber @xmath18 for model ( 2 ) , when @xmath78 .
the eigenvalues are positive for @xmath79 ( where the imaginary part is nonzero ) and for @xmath80 ( where the imaginary part is zero ) .
the imaginary part of the eigenvalue vanishes at @xmath81 . in the blue region ,
both turing and hopf instability exist , and only turing instability exists in the red region .
( * b * ) the emergent wavenumber @xmath82 ( green , + ) , marginal wavenumber for the blue region in ( * a * ) @xmath22(blue ) , and marginal wavenumber for the red region in ( * a * ) @xmath23 ( red ) are plotted as a function of @xmath24 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * sf2 * stationary pattern @xmath83 of the reaction - diffusion ( 5 ) and the predicted pattern of @xmath84 obtained from spatial map ( 4 ) .
the predicted pattern ( given by @xmath73 ) agrees well with the stationary pattern ( lines ) .
three patterns with boundary values of @xmath74 ( purple ) , @xmath85 ( green ) , and @xmath86 ( yellow ) are considered .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ | boundary - induced pattern formation from a spatially uniform state is investigated using one - dimensional reaction - diffusion equations .
the temporal oscillation is successively transformed into a spatially periodic pattern , triggered by diffusion from the fixed boundary .
we introduced a spatial map , whose temporal sequence , under selection criteria from multiple stationary solutions , can completely reproduce the emergent pattern , by replacing the time with space .
the relationship of the pattern wavelength with the period of oscillation is also obtained .
the generality of the pattern selection process and algorithm is discussed with possible relevance to biological morphogenesis .
spatial patterns are ubiquitous in non - equilibrium systems , and the process of spontaneous pattern formation has been extensively studied as one of the main issues in non - linear dynamics .
topic areas in pattern formation include fluid , solid - state , optical , geophysical , chemical , and biological systems@xcite .
in particular , the reaction - diffusion system , which was pioneered by turing@xcite , has been a focus of non - equilibrium studies . in his celebrated study ,
turing showed how a spatial pattern or temporal rhythm is spontaneously formed from a spatially homogeneous and temporally stationary state , which is unstable against perturbations .
turing classified such pattern formation processes into six cases , one of which is known as the _ turing pattern _ ; stationary periodic pattern with the finite wavelength given by linear stability analysis .
while turing originally proposed his theory as a model of morphogenesis , it has been applied to general pattern formation dynamics beyond developmental biology@xcite .
spontaneous pattern formation from temporally dynamic states is also possible in a reaction - diffusion system .
destabilization of a spatially homogeneous and temporally oscillatory state by diffusive interaction often results in spatiotemporal dynamics such as waves , spirals , and turbulence@xcite .
spatiotemporal dynamics by combination of turing instability for pattern formation and hopf bifurcation for temporal oscillation have been studied as turing - hopf bifurcation@xcite .
however , an important factor that has not been fully explored is the influence of boundary conditions on spatial patterns . in particular ,
pattern formation from temporally dynamic states may be crucially influenced by the introduction of a fixed boundary .
the influence may be globally propagated to alter the pattern dynamics .
( see also @xcite for the relevance of boundary conditions in pattern dynamics . ) in the present letter , we are concerned with such boundary - induced pattern formation , namely , how a fixed boundary condition destabilizes the stable , temporally periodic , spatially uniform state and transforms it into a temporally stationary , spatially periodic pattern .
the question of boundary - induced pattern formation is not only of theoretical interest but also of experimental interest in developmental biology . in the somitogenesis of vertebrate development , temporal oscillation in protein expression
is fixed into a spatial pattern@xcite . as
development progresses , the intracellular concentration of the protein c - hairy1 , which initially oscillates in the system , is fixed into a striped pattern along aligned cells .
so far , such spatial pattern formation has been studied by introducing external inputs that move in space with time development ( _ clock and wavefront_)@xcite .
diffusive interaction has been introduced by meinhardt in addition to external input using a spatial gradient@xcite . for these theoretical studies ,
the use of an external input is essential to destabilize the homogeneous state for pattern formation .
some recent experiments , however , suggested that an external input may not be essential , and intrinsic instability due to diffusion might result in somitogenesis@xcite .
if boundary - induced pattern formation is possible without imposing an external input throughout the space , it will provide a plausible mechanism for vertebrate somitogenesis . here
, we demonstrate that temporal oscillation in a one - dimensional reaction - diffusion system is fixed into a stationary periodic pattern by introducing fixed boundary condition .
in contrast to the celebrated turing pattern , the wavelength of the generated pattern can not be obtained using linear stability analysis around the fixed point .
instead , to predict the selected pattern , we introduced a one - dimensional spatial map , whose attractor gives the one - dimensional pattern by replacing time with space .
the generality of this oscillation fixation and the pattern selection mechanism will be discussed .
now , consider a one - dimensional reaction - diffusion system of two components @xmath0 and @xmath1 .
we assume that the diffusion of @xmath0 is much faster than that of @xmath1 , and the diffusion of the latter is neglected for simplicity ( the formalism to be discussed is valid even without this approximation ) .
then , the equation is written as @xmath2\ \dfrac{dy}{dt}=g(x , y)\\[12pt ] \end{array } \right .
\label{eq : reaction - diffusion}\ ] ] where @xmath3 and @xmath4 are the reaction functions for @xmath0 and @xmath1 , @xmath5 is the diffusion constant of @xmath0 , and the attractor of the dynamical system without @xmath5 is a limit cycle , the eigenvalues of the jacobi matrix around the fixed point satisfy the positive real part with a non - zero imaginary part so that @xmath6 and @xmath7@xcite , where @xmath8 ] [ eq : jacobian ] .
we consider the case in which the spatially uniform , limit cycle state is stable against perturbations so that , under neumann or periodic boundary conditions , this state is the attractor .
however , under a fixed boundary condition , the variable @xmath0 close to the boundary can not oscillate , which may destabilize the oscillatory attractor .
indeed , we have found that a fixed periodic pattern is often generated in this case . as a specific example , consider the following system , @xmath9 g(x , y)=\dfrac{1}{1+e^{-\beta ( y - x)}}-y\\ \end{array } \right .
\label{eq : example1}\ ] ] which describe the protein expression dynamics with two genes , where @xmath0 inhibits the expression of @xmath1 and @xmath1 activates the expressions of both @xmath0 and @xmath1@xcite . without the diffusion term
, this system has one unstable fixed point @xmath10 if @xmath11 , and the limit cycle is an attractor .
indeed , from initial conditions close to a spatially homogenous state , a uniform , limit - cycle state is reached if the boundary condition is neumann or periodic .
in contrast , if a fixed boundary condition is adopted for at least one end , i.e. , @xmath12 , temporal oscillation is replaced by a fixed , spatially periodic pattern is not required as diffusion in @xmath1 is neglected . ]
( see fig . 1a and 1b where @xmath13 is adopted at the other end ) .
note that a pattern of the same wavelength is organized independently of @xmath14 as long as @xmath14 is between [ 0,1](see fig .
3 ) . here
, oscillation ceased in the vicinity of @xmath15 , which works as a boundary for @xmath16 for slightly larger @xmath17 .
the oscillation is successively fixed for larger @xmath17 .
we analytically examined how the organized pattern is determined .
first , we confirmed that the linear stability analysis around the uniform fixed - point solution can not explain the wavelength in contrast to the turing instability case .
the real part of eigenvalue is positive against perturbations of a given wavenumber @xmath18 around the unstable fixed point solution @xmath19 for @xmath20 , while the imaginary part is nonzero . on the other hand , for @xmath21 , one of the eigenvalues is positive with a vanishing imaginary part where turing instability exists .
thus , both hopf and turing instability coexist in this system ( see supplemental figure 1a ) .
then , we contrasted the marginal values @xmath22 and @xmath23 with the wavenumber of the emergent pattern by changing @xmath24 .
the observed wavelength agrees neither with @xmath22 nor with @xmath23(see supplemental figure 1b ) .
even the dependence of the parameters on @xmath24 does not agree .
indeed , this discrepancy is natural : pattern formation occurs from the homogeneous limit cycle , which is far from the unstable fixed solution .
hence , the standard analysis for the turing pattern does not work in this system .
now we need to find a procedure to determine the spatial pattern without utilizing linear stability analysis .
we note that once the diffusion term is given , the stationary solution @xmath25 can be obtained from the dynamical system of two variables @xmath26 , where standard nullcline analysis works . as shown in fig .
2b , only the nullcline of @xmath0 ( not that of @xmath1 ) is horizontally shifted with the diffusion term .
accordingly , the fixed point ( @xmath27 is shifted in the state space .
once the fixed point @xmath25 is given , the gradient term is obtained . for a fixed pattern , the fixed point @xmath28 and the gradient term
are determined self - consistently .
this self - consistent condition , given by @xmath29 , can be solved explicitly by adopting spatial discretization to compute the diffusion term . in general , this condition is given by @xmath30 \end{array } \right .
\label{eq : discretized_equation}\ ] ] where @xmath31 is discretized space index with @xmath32 as the discretized unit length , i.e. , @xmath33 and @xmath34 , @xmath35 . here
, @xmath36 is the rescaled diffusion constant which satisfies @xmath37 . from these equations
, we can derive the spatial map as follows : @xmath38 where @xmath39 is the inverse function of @xmath4 given that @xmath0 is fixed to @xmath40 , which corresponds to the cross - points of the nullclines . with these equations , @xmath41 is determined from @xmath40 , @xmath42 , and @xmath43 , while @xmath44 is determined as @xmath45 ( for the use of a `` spatial map '' in the analysis of a spatial pattern , see also @xcite ) .
however , there can be multiple candidate solutions for @xmath45 .
in fact , not all of the above solutions are stable in terms of dynamical systems ( [ eq : reaction - diffusion ] ) .
furthermore , this solution must be attracted from the uniform oscillatory state .
these properties lead to the following selection principles .
+ + * selection principles * 1 .
ignore candidates in which the resulting state @xmath46 is not stable against small perturbations in @xmath41 .
2 . ignore candidates in which the resulting value @xmath47 is out of the basin of the original limit cycle .
3 . calculate @xmath47 from all remaining candidates , and choose the solution that gives the smallest value of @xmath48 .
the first criterion is necessary for the stability of the novel fixed point .
the second criterion is necessary for the pattern to emerge from the limit cycle .
the third criterion implies the smallest net diffusion , which minimizes the distance from a uniform state . with the second and third criteria , the solution that is attracted from the uniform oscillatory state
is selected .
an example of the pattern obtained from the spatial map with the above selection criteria is shown in fig .
3 , which agrees well with the numerically obtained pattern .
the above procedure works independently of the parameter values , demonstrating its validity .
note that the spatial periodic pattern is obtained as an attractor of the spatial map ( [ eq : spatialmap ] ) .
hence , independently of the initial condition in the map , the same periodic pattern is selected as long as the initial condition in the spatial map belongs to the basin of the above attractor .
the initial condition in the spatial map corresponds to the value of fixed boundary .
hence , a pattern of the same wavelength is reached independently of the boundary value , while the phase of wave pattern is different , which is consistent with the numerical result ( see fig .
3 ) . in the state space ,
pattern selection using the spatial map can be described as follows .
when the diffusion term is small , the shift of the nullcline is small .
the same stable fixed point , i.e. , @xmath43 on the same branch , continues to exist and is selected according to the above criteria .
with iteration of the spatial map , the shift of the nullcline due to the diffusion term is accumulated and at some point , the fixed point of the same branch vanishes with bifurcation , or do not fulfill selection principles .
the selected fixed point is replaced by that of the new branch , as the solution according to the above criteria . repeating the above procedure
, a periodic pattern is obtained .
this procedure suggests there can be a relationship between the period of oscillation @xmath49 and the wavelength @xmath50 .
first , by rescaling the spatial scale , the wavelength @xmath50 is proportional to @xmath51 .
( the period is scaled by @xmath52 if @xmath53 and @xmath54 are similarly scaled by @xmath55 , but this is already set to unity in ( [ eq : reaction - diffusion ] ) . ) now , following the vector field @xmath56 in the state space ( fig .
2a ) , @xmath57 oscillates with the period @xmath49 . on the other hand , when a stripe is formed , the `` spatial orbit '' @xmath16 along space @xmath17 orbits the state space similarly to the limit cycle orbit .
indeed , except for the vicinity of the hopf bifurcation , we numerically confirmed that one stripe is generated in correspondence with one period of oscillation . with the parameter change , the ( magnitude of ) vector @xmath58 in the state space changes , which changes the rate of change along the limit cycle as well as that of the `` spatial orbit '' in the same way .
if these orbits do not have a small radius in the state space , both speeds are expected to change in proportion , implying that the period @xmath49 and wavelength @xmath50 ( or @xmath59 ) change in proportion .
of course , this is a rough estimate , but this proportionality approximately holds against changes in the parameter values @xmath24 ( and @xmath5 ) , as shown in fig .
4 .
the fixation of the temporal to the spatially periodic pattern as well as the analysis using the spatial map can be generally applied to a one - dimensional reaction - diffusion equation that has a uniform limit - cycle attractor . as another illustration
, we examined the so - called brusselator@xcite : @xmath60 g(x , y)=a-(b+1)x+xy^2\\ \end{array } \label{eq : brusselator}\ ] ] in this case , a uniform periodic oscillation is replaced by a periodic spatial pattern by applying the fixed boundary condition @xmath12 for a certain range of parameters , which agrees with the prediction using the corresponding spatial map ( see supplemental figure 2 ) .
the approximate proportionality between the period and wavelength is worse than the model ( 2 ) ( see fig .
4 ) , possibly because the change in the flow @xmath61 against the parameters is highly dependent on the state @xmath62 . in the present letter
, we have studied the formation of a periodic pattern from a uniform oscillatory state induced by the boundary conditions .
the emergent pattern is predicted as an attractor of the spatial map under the selection principle .
we have confirmed the generality of this pattern formation in reaction - diffusion systems .
the periodic oscillation exists in a two - component reaction system with negative feedback .
that is , in a system with an activator and an inhibitor , for a certain range of parameters , the spatially periodic pattern that exists as an attractor replaces the uniform oscillatory state , triggered by the fixed boundary condition , if the diffusion of the inhibitor is sufficiently large .
we also note that the pattern formation process as well as the nullcline analysis is still valid even for @xmath63 , although the spatial map explicitly includes the inverse function of @xmath1 , which would make the analysis more difficult
. pattern formation generally depends on the boundary and initial conditions . in the spatial map ,
the former is represented by the initial condition in the map , while the latter is considered by the second and third criteria of the selection principle , which indicate that the initial condition is not far from the uniform oscillatory state .
if the initial condition is far from uniformity , local inhomogeneity can grow . in the state - space representation , whether such growth occurs is determined according to the size of the spatial diffusion term , specifically whether the spatial diffusion term is large enough to go across the nullclines and to induce saddle - node bifurcation .
in other words , the condition close to uniformity represents the absence of such inhomogeneity . on the other hand , for such initial conditions to allow for the local growth in inhomogeneity , the selection criteria are replaced so that the choice of the fixed - point solution is close to a given initial condition , leading to a requested switch to a different branch of fixed - points . hence , the choice of the selection principle corresponding to the initial condition can predict the emergent pattern .
control of an emergent pattern is thus possible by manipulating the initial condition . here
, one stripe is formed with one - to - one correspondence of one oscillation period in time . near the hopf bifurcation point , however , a complex pattern with one stripe is observed per few periods of oscillation .
indeed , the spatial map can have an attractor with such complex ( or quasiperiodic ) oscillation in the vicinity of the bifurcation and also by modifying the selection principle .
analysis of such complex patterns in terms of the spatial map ( see also @xcite ) will be of interest in the future .
experimental confirmation of the present pattern formation will be possible in reaction - diffusion systems .
in particular , relevance of boundary condition to pattern selection should be of importance because the real experimental system is finite and often under a fixed boundary condition . by carefully examining the boundary effects
, we can confirm the present pattern formation mechanism , where the period - wavelength relationship ( fig .
4 ) will be confirmed . as mentioned in the introduction , spatial pattern formation based on temporal oscillation
is often observed in biological morphogenesis@xcite as well as in the numerical evolution of morphogenesis@xcite , where the relevance of cell - cell interaction has been recently discussed in addition to the external morphogen gradient@xcite .
considering the simplicity in our mechanism , which only requires diffusion of the inhibitor and a fixed boundary , we expect that it could be adopted in biological development , which will be confirmed by examining cell - cell interaction and boundary effects .
the authors would like to thank nen saito for useful discussions .
this work was partially supported by a grant - in - aid for scientic research ( no .
21120004 ) on innovative areas neural creativity for communication " ( no .
4103 ) and the platform for dynamic approaches to living systems from mext , japan .
28ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _ _ ( , ) @noop * * , ( ) @noop _ _ ( , ) @noop _ _ , vol .
( , ) @noop _ _ , vol .
( , ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1126/science.251.4994.650 [ * * , ( ) ] @noop _ _ , vol .
( , ) link:\doibase 10.1103/physreve.54.261 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , vol .
( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physreve.88.032718 [ * * , ( ) ] in @noop _ _ ( , ) pp . @noop * * , ( ) link:\doibase 10.1371/journal.pone.0002772 [ * * , ( ) ] @noop * * , ( ) |
the electroweak standard model ( ewsm ) is in good agreement with all experimentally known phenomena of electroweak origin , with the exception of the evidence of neutrino mixing .
the only ingredient predicted by this model that has not been seen yet is the higgs particle .
the direct search at lep provided us with a lower limit on the mass of the standard model higgs - boson @xmath3 excluding the region below @xmath4gev @xcite .
the global fit of the experimental data to the standard model , based on confronting theoretical predictions for electroweak observables with their experimental values , favours a light higgs - boson ( @xmath5gev one - sided @xmath6 confidence level ( cl ) @xcite ) .
+ however , there is a discrepancy of about @xmath7 between the two most precise measurements of the electroweak mixing angle @xmath1 , which is used to set stringent bounds on the higgs - boson mass . the measurement based on the leptonic asymmetry parameter @xmath8 at sld , together with the @xmath2-boson mass measurement from tevatron and lep , point to a light higgs - boson with a mass slightly below the lower bound from the direct searches . on the other hand , the measurement based on the @xmath9-quark forward - backward asymmetry @xmath10 at lep favours a relatively heavy higgs - boson with a mass around @xmath11gev @xcite .
since the center of mass energies of present colliders do not allow to probe the region of a heavy higgs - boson , the sensitivity of radiative corrections to low energy electroweak observables to a heavy - higgs boson mass becomes an important tool in setting limits on @xmath3 @xcite .
+ as the yukawa couplings are very small , the higgs dependent effects are limited to corrections to the vector - boson propagators .
they give rise to shifts , e.g. , in the @xmath0-parameter , the @xmath2-boson mass , and in the effective leptonic weak mixing angle @xmath1 .
these shifts are often parametrized by s , t and u or @xmath12 @xcite . + for a light higgs - boson , the higgs mass dependence of theoretical predictions is mainly due to one - loop radiative corrections to the gauge - boson propagators which grow logarithmically with @xmath3 @xcite .
however , because the higgs self - interaction is proportional to @xmath13 , higher order radiative corrections which grow like powers of @xmath3 could become important if the higgs - boson mass is much larger than the z - boson mass . + at the two - loop level , the leading corrections are proportional to @xmath13 , but the numerical coefficient of these terms turns out to be very small @xcite , and therefore they are not important for @xmath3 less than a few tev .
however it has been suggested that the smallness of the two - loop corrections may be somewhat accidental @xcite , therefore important effects might first appear only at the three - loop level .
+ the situation was only clarified recently by an explicit leading three - loop calculation for three precision variables , the @xmath0-parameter , the electroweak mixing angle @xmath1 , and the @xmath2-boson mass @xmath14 @xcite .
+ the electroweak @xmath0-parameter is a measure of the relative strengths of neutral and charged - current interactions in four - fermion processes at zero momentum transfer . in the standard model , at tree level , it is related to the w and z boson masses by : @xmath15 where @xmath16 . including higher order corrections modifies this relation into @xmath17 here @xmath18 parametrises all higher loop corrections which are sensitive to the existence of a heavy higgs particle . the leading one- and two - loop corrections , which grow logarithmically and quadratically with @xmath3 respectively , have been calculated @xcite . + the sine of the effective leptonic weak mixing angle @xmath1 is defined in terms of the couplings of the @xmath19-boson to leptons .
the complete electroweak fermionic corrections to @xmath1 at the two - loop level are known @xcite .
recently , the higgs - dependent electroweak two - loop bosonic contributions to this observable have been completed @xcite . for the @xmath2-boson mass , both the fermionic and the bosonic corrections have been calculated at the two - loop level @xcite .
+ in this contribution we review the leading three - loop bosonic corrections to the @xmath0-parameter , @xmath1 and @xmath14 @xcite , which grow like @xmath20 in the large higgs mass limit . +
the calculation is organized in such a way that the leading contributions come only from self energy corrections to the gauge boson propagators , often referred to as _
oblique _ corrections in the literature , and not from vertex or box diagrams .
this is achieved by our choice of renormalization scheme .
we should mention at this point that the renormalization is performed up to the two - loop level only , removing sub - divergences from the gauge boson self - energies , but not yet the overall divergences .
this is due to the fact that the three - loop counter terms cancel in @xmath21 , @xmath22 and @xmath23 .
the renormalized self - energies are then related to physical observables through the formalism of s , t and u parameters , which was developed by peskin and takeuchi to describe the effect of heavy particles on electroweak precision observables @xcite .
they are defined in terms of the transverse gauge boson self - energies at zero momentum transfer and their first derivative w.r.t their momentum .
these self - energies are not observable individually and may still contain ultraviolet divergences .
however the three combinations : @xmath24 where @xmath25 and @xmath26 is the fine structure constant , are finite and observable . using this formalism
, the shifts to @xmath27 and @xmath14 are parametrised as : @xmath28
the bare gauge boson self - energies are decomposed into transversal and longitudinal components according to @xmath29 where @xmath30 .
the scalar functions @xmath31 are then expanded in a taylor series in their momentum @xmath32 up to order @xmath33 .
higher order terms in @xmath33 are suppressed by powers of the heavy higgs boson mass . for the @xmath34 parameter ,
only the constant term of this expansion ( i.e @xmath35 ) is required .
after this step , we are left with vacuum integrals which , in general , depend on three different scales : @xmath36 and @xmath37 . in order to separate the dependence of these integrals on the large scale @xmath3 from the small scales @xmath38 and @xmath37 and extract the leading @xmath3 terms , we perform an asymptotic large mass expansion following the method of the expansion by regions @xcite .
the expansion is constructed by considering different regions in loop momentum space , distinguished by the set of propagator momenta which are large or small in those regions . in each region
, a taylor expansion of all propagators in the small masses and in the small momenta of that region is performed .
typically , the expansion generates extra scalar products of loop momenta in the numerator and higher powers of denominators , as compared to the original diagrams .
the resulting expression is then integrated over the whole loop momentum space .
for the three - loop vacuum topology shown in fig [ mercedes ] , there are @xmath39 regions in loop momentum space to consider ( see @xcite for more details ) .
the distribution of large and small masses in each diagram decides the number of regions that contribute to the corresponding integral up to the leading terms we are interested in .
for example , only four regions give a non vanishing contribution to the diagram shown in fig [ wwexpregexample ] , up to @xmath20 order .
they correspond to : * the all internal momenta large region . in this case one
expands the propagators in @xmath40 only , the result is a one - scale three - loop integral .
* the region where @xmath41 is small .
here we expand in @xmath42 and @xmath41 , which leads to the product of a one - loop integral depending on @xmath40 times a two - loop integral that depends on @xmath3 . * the region where @xmath43 is small .
after expanding in @xmath40 and @xmath43 we get , similarly to the previous case , a product of one- and two - loop integrals . * the region where @xmath44 is small . here
again we expand in @xmath42 and @xmath44 and get a product of one- and two loop integrals .
all the other regions produce either scaleless integrals , which are zero in dimensional regularization , or terms which do not have enough powers of @xmath3 to contribute to the leading order .
( 80,80)(0,0 ) ( 50,50)(40,0,180 ) ( 50,50)(40,180,360 ) ( 50,90)1.5 ( 50,50)1.5 ( 50,90)(50,50 ) ( 16,30)1.5 ( 16,30)(50,50 ) ( 84,30)1.5 ( 84,30)(50,50 ) [ mercedes ] from the expansion in the @xmath39 regions we get two kinds of integrals : * factorizable diagrams , which are products of one-(two ) loop vacuum integrals depending on @xmath3 and two-(one- ) loop vacuum integrals depending on @xmath14 and @xmath37 .
* non - factorizable three - loop vacuum integrals , which are either single - scaled depending on @xmath3 , or double - scaled depending on @xmath14 and @xmath37 .
( 110,110)(0,25 ) ( 100,100)301 ( 100,100)(100,130 ) ( 75,84)(100,100 ) ( 100,100)(125,84 ) ( 60,37)(90,72.5)26 ( 90,72.5)1 ( 140,37)(110,72.5)26 ( 110,72.5)1 ( 59,30 ) [ ] ( 141,30 ) [ ] ( 100,64 ) [ ] ( 74,73.5)[]@xmath45 ( 128,73.5)[]@xmath45 ( 78,133)[]@xmath45 ( 128,133)[]@xmath45 ( 94,115 ) [ ] ( 80,95 ) [ ] ( 120,95 ) [ ] ( 100,74 ) [ ] ( 90,79 ) [ ] ( 110,79 ) [ ] ( 91,90 ) [ ] ( 107,90 ) [ ] ( 106,115 ) [ ] ( 150,116 ) [ ] ( 58,116 ) [ ] ( 82,45 ) [ ] ( 68,38)(80,52 ) ( 120,52)(132,38 ) [ wwexpregexample ] the integration - by - parts ( ibp ) method @xcite is then used to reduce all the vacuum integrals to the master ones .
we classify the single - scaled three - loop non - factorizable integrals into ten different kinds , depending on the distribution of masses in the propagators .
their reduction to a small set of master integrals @xcite was done in two ways . on the one hand ,
reduction formulae based on the integration by part identities have been constructed .
on the other hand the automatic integral reduction package @xmath46 @xcite was used as a cross check .
the latter was also used to reduce the double - scaled three - loop non - factorizable integrals .
explicit formulae for their master integrals are not needed , as they all canceled once we have summed over all diagrams .
+ the longitudinal parts of the gauge boson self - energies are related to the self - energies of the goldstones and mixings between gauge bosons and goldstones by a set of ward identities .
we have verified that these ward identities are satisfied by the full ( i.e including tadpoles ) unrenormalised self - energies up to order @xmath33 . as we have mentioned earlier ,
leaving out vertex and box contributions requires a proper way of renormalizing .
our renormalization conditions are fixed in such a way that the renormalization removes all the terms of order @xmath13 and @xmath20 from the one- and two - loop gauge boson self - energies , the charged and neutral goldstone self - energies and the mixings between gauge bosons and goldstones .
this ensures that no two- or three - loop vertex or box graphs containing such self - energies as subgraphs can give corrections that grow like @xmath13 or @xmath20 in the large higgs mass limit ( see fig . [ vertexzzse ] ) .
( 100,70)(0,-5 ) ( 30,0)1.5 ( 30,60)1.5 ( 60,30)1.5 ( 85,30)1.5 ( 100,0)(85,30 ) ( 85,30)(100,60 ) ( 0,0)(30,0 ) ( 30,0)(60,30 ) ( 60,30)(30,60 ) ( 30,60)(0,60 ) ( 60,30)(85,30)35 ( 30,20)1.5 ( 30,40)1.5 ( 30,60)(30,40)-33.5 ( 30,0)(30,20)33.5 ( 30,30)10midnightbluegray ( 98,-4 ) [ ] ( 98,63 ) [ ] ( 2,-6.5 ) [ ] ( 2,66 ) [ ] ( 22,50 ) [ ] ( 22,12 ) [ ] ( 29,29 ) [ ] ( 72,22 ) [ ] [ vertexzzse ] as a check on the renormalization , we have verified that the renormalized longitudinal photon self - energy and photon - z mixing are zero .
the shifts to the electroweak precision observables @xmath0 , @xmath1 and @xmath14 relative to their tree level values , expressed in terms of @xmath47 and @xmath37 , are given by @xmath48 with @xmath18 , @xmath49 and @xmath50 defined in ( [ eq : deltasin ] ) in terms of the parameters @xmath51 , @xmath34 and @xmath52 .
while the @xmath52-parameter vanishes in the approximation where only quartic terms or higher powers in @xmath3 are kept at the three - loop level , the @xmath51- and the @xmath34-parameters give the following contributions @xmath53 using @xmath54 for the weak coupling constant , with and @xmath55 , the shifts to @xmath0 , @xmath1 and @xmath14 are @xmath56 ( 0,0 ) as a function of @xmath57._,title="fig : " ] ( 13500,8100)(0,0)(2750,1650 ) ( 0,0)[r ] -0.007 ( 2750,2493 ) ( 0,0)[r ] -0.006 ( 2750,3336 ) ( 0,0)[r ] -0.005 ( 2750,4179 ) ( 0,0)[r ] -0.004 ( 2750,5021 ) ( 0,0)[r ] -0.003 ( 2750,5864 ) ( 0,0)[r ] -0.002 ( 2750,6707 ) ( 0,0)[r ] -0.001 ( 2750,7550 ) ( 0,0)[r ] 0.000 ( 4284,1100 ) ( 0,0 ) 5 ( 6382,1100 ) ( 0,0 ) 10 ( 8479,1100 ) ( 0,0 ) 15 ( 10577,1100 ) ( 0,0 ) 20 ( 12675,1100 ) ( 0,0 ) 25 ( 550,4600)(7850,275 ) ( 0,0 ) @xmath57 ( 7575,4010 ) ( 0,0)[r ] @xmath58 ( 7575,3448 ) ( 0,0)[r ] @xmath59 ( 7575,2886 ) ( 0,0)[r ] @xmath60 [ fig : rho ] ( 0,0 ) as a function of @xmath57._,title="fig : " ] ( 13500,8100)(0,0)(1925,1650 ) ( 0,0)[r ] 0.0 ( 1925,2388 ) ( 0,0)[r ] 0.5 ( 1925,3125 ) ( 0,0)[r ] 1.0 ( 1925,3863 ) ( 0,0)[r ] 1.5 ( 1925,4600 ) ( 0,0)[r ] 2.0 ( 1925,5338 ) ( 0,0)[r ] 2.5 ( 1925,6075 ) ( 0,0)[r ] 3.0 ( 1925,6813 ) ( 0,0)[r ] 3.5 ( 1925,7550 ) ( 0,0)[r ] 4.0 ( 3566,1100 ) ( 0,0 ) 5 ( 5843,1100 ) ( 0,0 ) 10 ( 8121,1100 ) ( 0,0 ) 15 ( 10398,1100 ) ( 0,0 ) 20 ( 12675,1100 ) ( 0,0 ) 25 ( 550,4600)(7437,275 ) ( 0,0 ) @xmath57 ( 7163,6370 ) ( 0,0)[r ] @xmath58 ( 7163,5808 ) ( 0,0)[r ] @xmath59 ( 7163,5246 ) ( 0,0)[r ]
@xmath60 [ fig : sinth ] the higgs mass dependence of the @xmath0-parameter and @xmath49 is shown in figs .
[ fig : rho ] and [ fig : sinth ] .
it turns out that the sign of the leading three - loop corrections to @xmath18 , @xmath49 and @xmath50 is the same as the sign of the one - loop contributions .
the original question that motivated these calculations was , whether inclusion of the three - loop corrections with strong interactions could lead to an effect mimicking the one - loop effects of a light higgs boson .
the result of the investigations shows that this is highly unlikely .
as the signs of the three - loop corrections are the same as the ones of the one - loop corrections , with increasing higgs mass , the three - loop terms only make the effects grow faster , instead of partially cancelling the one - loop corrections
. therefore the presence of a strongly interacting heavy higgs - sector appears to be extremely unlikely , and the electroweak precision data can indeed be taken as a strong indication for a light higgs boson sector .
r. b. would like to thank the organizers of the conference for the pleasant atmosphere .
this work was supported by the sofja kovalevskaja award of the alexander von humboldt foundation sponsored by the german federal ministry of education and research and the dfg within the project `` ( nicht)-perturbative quantenfeldtheorie '' .
r. barate _
[ aleph collaboration ] , phys .
b * 565 * ( 2003 ) 61 [ arxiv : hep - ex/0306033 ] .
the lep collaborations aleph , delphi , l3 and opal , the lep electroweak working group , the sld electroweak group and the sld heavy flavor group , arxiv : hep - ex/0509008 .
s. eidelman _ et al . _
[ particle data group ] , phys .
b * 592*,(2004 ) 1 .
m. j. g. veltman , acta phys .
b * 8 * ( 1977 ) 475 .
j. j. van der bij , nucl .
b * 248 * ( 1984 ) 141 ; + j. j. van der bij , nucl .
b * 267 * ( 1986 ) 557 .
j. van der bij and m. j. g. veltman , nucl .
b * 231 * ( 1984 ) 205 .
r. barbieri , p. ciafaloni and a. strumia , phys .
b * 317 * ( 1993 ) 381 .
r. akhoury , j. j. van der bij and h. wang , eur .
j. c * 20 * ( 2001 ) 497 [ arxiv : hep - ph/0010187 ] .
r. boughezal , j. b. tausk and j. j. van der bij , nucl .
b * 713 * ( 2005 ) 278 [ arxiv : hep - ph/0410216 ] ; + r. boughezal , j. b. tausk and j. j. van der bij , nucl .
b * 725 * ( 2005 ) 3 [ arxiv : hep - ph/0504092 ] m. awramik , m. czakon , a. freitas and g. weiglein , phys .
* 93 * ( 2004 ) 201805 [ arxiv : hep - ph/0407317 ] ; + w. hollik , u. meier and s. uccirati , arxiv : hep - ph/0507158 .
w. hollik , u. meier and s. uccirati , arxiv : hep - ph/0509302 .
m. awramik , m. czakon , a. freitas and g. weiglein , phys .
d * 69 * ( 2004 ) 053006 [ arxiv : hep - ph/0311148 ] .
m. e. peskin and t. takeuchi , phys .
* 65 * ( 1990 ) 964 ; + m. e. peskin and t. takeuchi , phys .
d * 46 * ( 1992 ) 381 .
v. a. smirnov , `` applied asymptotic expansions in momenta and masses , '' springer - verlag , berlin ( 2002 ) .
f. v. tkachov , phys .
* b100 * ( 1981 ) 65 ; + k. g. chetyrkin and f. v. tkachov , nucl .
* b192 * ( 1981 ) 159 . c. anastasiou and a. lazopoulos , jhep
* 0407 * , 046 ( 2004 ) [ arxiv : hep - ph/0404258 ] .
d. j. broadhurst , eur .
j. c * 8 * ( 1999 ) 311 [ arxiv : hep - th/9803091 ] .
j. fleischer and m. y. kalmykov , phys .
b * 470 * ( 1999 ) 168 [ arxiv : hep - ph/9910223 ] . | a review of the calculation of the leading three - loop electroweak corrections to the @xmath0-parameter , @xmath1 and the @xmath2-boson mass is presented .
the heavy higgs mass expansion and the renormalization are discussed .
.7 cm 3.5 cm 1.2 cm radja boughezal + _ institut fr theoretische physik und astrophysik , universtt wrzburg , , d-97074 wrzburg , germany _ + and + bas tausk + _ fakultt fr mathematik und physik , albert - ludwigs - universitt freiburg , + d-79104 freiburg , germany _ + 1.5 cm |
the theory of uniform distribution is concerned with the distribution of real numbers in the unit interval @xmath6 and it s development started with hermann weyl s celebrated paper @xcite .
this theory gives a useful technique for numerical calculation exactly of the one - dimensional riemann integral over @xmath0 $ ] .
more pricisely , the sequence of real numbers @xmath7^{\infty } $ ] is uniformly distributed in @xmath0 $ ] if and only if for every real - valued riemann integrable function @xmath8 on @xmath9 $ ] the equality @xmath10 holds ( see , for example @xcite , corollary 1.1 , p. 3 ) .
main corollaries of this assertion successfully were used in diophantine approximations and have applications to monte - carlo integration ( cf .
@xcite,@xcite,@xcite,@xcite ) .
note that the set @xmath11 of all uniformly distributed sequences in @xmath0 $ ] viewed as a subset of @xmath0^{\infty}$ ] has full @xmath12-measure , where @xmath12 denotes the infinite power of the linear lebesgue measure @xmath2 in @xmath0 $ ] ( cf .
@xcite , theorem 2.2(hlawka ) , p. 183 ) . for a fixed lebesgue integrable function @xmath8 in
@xmath0 $ ] , one can state a question asking _ what is a maximal subset @xmath13 of @xmath11 each element of which can be used for calculation it s lebesgue integral over @xmath0 $ ] by the formula ( 1.1 ) and whether this subset has the full @xmath12-measure_. this question has been resolved positively by kolmogorov s strong law of large numbers . there
naturally arises another question asking whether can be developed analogous methodology for invariant extensions of the lebesgue measure in @xmath0 $ ] and whether main results of the uniform distribution theory will be preserved in such a situation . in the present manuscript
we consider this question for a certain strictly separated defined on the measure space @xmath14 is called strictly separated if there exists a partition @xmath15 of the set @xmath16 such that @xmath17 for @xmath18 . ]
maximal ( in the sense of cardinality ) family of invariant extensions of the linear lebesgue measure in @xmath0 $ ] . in our investigations
we essentially use the methodology developed in works @xcite , @xcite , @xcite . the rest of the present paper it the following .
in section 2 we consider some auxiliary facts from the theory of invariant extensions of the lebesgue measure and from the probability theory . in section 3
we present our main results .
section 4 presents historical background of the theory of invariant extensions of the @xmath19-dimensional lebesgue measure as well haar measure in locally compact hausdorff topological groups . in section 5 we state a uniform distribution problem for invariant extensions of the haar measure in locally compact hausdorff topological groups
as usual , we denote by @xmath20 the real axis with usual metric and addition @xmath21 operation under which @xmath20 stands a locally compact @xmath22-compact hausdorff topological group .
we denote by @xmath2 the linear lebesgue measure in @xmath20 .
( @xcite , lemma 6 , p. 174 ) let @xmath23 be a shift - invariant @xmath22-ideal of subsets of the real axis @xmath20 such that @xmath24 where @xmath25 denotes an inner measure defined by the linear lebesgue measure @xmath2 .
then the functional @xmath26 defined by @xmath27 where @xmath28 is lebesgue measurable subset of @xmath20 and @xmath29 and @xmath30are elements of the @xmath22-ideal @xmath23 , is a shift - invariant extension of the lebesgue measure @xmath2 . for the proof of lemma 2.1 , see , e.g. , @xcite , @xcite .
( @xcite , lemma 4 , p. 164 ) there exists a family @xmath31}$ ] of subsets of the real axis @xmath20 such that : @xmath32 \ \ & \ i ' \in [ 0,1 ] \ \ & \ i \neq i ' \rightarrow
x_i \cap x_{i'}=\emptyset ) ; $ ] @xmath33 ~\&~ ( f$ ] is a closed subset of the real axis @xmath20 with @xmath34 @xmath35 \ \ & \ g \in r \rightarrow card(g+(\bigcup_{i \in i ' } x_i ) \triangle ( \bigcup_{i \in i ' } x_i ) ) < c ) .$ ] there exists a family @xmath36}$ ] of measures defined on some shift - invariant @xmath22-algebra @xmath37 of subsets of the real axis @xmath20 such that : @xmath38 \rightarrow$ ] the measure @xmath39 is a shift - invariant extension of the linear lebesgue measure @xmath40 @xmath41 \ \ & \ t ' \in [ 0,1 ] \ \ & \ t \neq t ' \rightarrow \mu_t$ ] and @xmath42 are orthogonal and @xmath42 are called orthogonal if there exists @xmath43 such that @xmath44 and @xmath45 measures .
moreover , @xmath46 for each @xmath47 $ ] , where @xmath48}$ ] comes from lemma 2.2 .
@xmath49 for arbitrary @xmath47 $ ] , we denote by @xmath50 an shift - invariant @xmath22-ideal generated by the set @xmath51 . then it is easy to verify that the @xmath22-ideal @xmath50 satisfies all conditions of lemma 2.1 .
let us denote by @xmath52 the shift - invariant extension of the lebesgue measure @xmath2 produced by the @xmath22-ideal @xmath50 .
we obtain the family @xmath53}$ ] of shift - invariant extensions of the lebesgue measure @xmath2 .
denote by @xmath37 the shift - invariant @xmath22-algebra of subsets of the real axis @xmath20 , generated by the union @xmath54 \},\ ] ] where @xmath55 and @xmath56 denotes the class of all lebesgue measurable subsets of the real axis @xmath20 .
also , assume that @xmath57 \rightarrow \mu_t=\overline{\mu}_t|_{s(r)}).\ ] ] if we consider the family of shift - invariant measures @xmath36 } $ ] , we can easily conclude that this family satisfies all conditions of lemma 2.3 .
let consider the family @xmath36}$ ] of shift - invariant extensions of the measure @xmath2 which comes from lemma 2.3 .
let denote by @xmath4 the restriction of the measure @xmath39 to the class @xmath58:=\ { y \cap [ 0,1 ] : y \in s(r ) \},\ ] ] where @xmath37 comes from lemma 2.3 .
it is obvious that for each @xmath47 $ ] , the measure @xmath4 is concentrated on the set @xmath59 $ ] provided that @xmath60 \setminus c_t)=0 $ ] .
the next proposition is useful for our further consideration .
( kolmogorov strong law of large numbers , @xcite , theorem 3 , p.379 ) let @xmath61 be a probability space and @xmath62 be a sequence of independent equally distributed random variables for which mathematical expectation @xmath63 of @xmath64 is finite
. then the following condition @xmath65 holds .
let consider the family of probability measures @xmath1}$ ] and the family @xmath66}$ ] of subsets of @xmath0 $ ] which come from remark 2.4 . for @xmath47 $ ]
, we denote by @xmath67,\lambda_t)$ ] the class of @xmath4-integrable functions .
then for @xmath68,\lambda_t)$ ] , we have @xmath69^{\infty } ~\&~ \lim_{n \to \infty}\frac{\sum_{k=1}^nf(x_k)}{n}=\int_{[0,1 ] } f(x)d \lambda_t(x)\})=1.\ ] ] for fixed @xmath47 $ ] , we set @xmath70 where @xmath71\},$ ] where @xmath72 $ ] comes from remark 2.4 .
@xmath73 , where @xmath74 denotes restriction of the measure @xmath4 to the sigma algebra @xmath75 . for @xmath76 and
@xmath77 we put @xmath78
. then all conditions of lemma 2.5 are satisfied which implies that @xmath79 @xmath80 equivalently , @xmath81 the latter relation implies @xmath69^{\infty } ~\&~ \lim_{n \to \infty}\frac{\sum_{k=1}^nf(x_k)}{n}=\int_{[0,1 ] } f(x)d \lambda_t(x)\})\ge\ ] ] @xmath81 a sequence of real numbers @xmath82^{\infty}$ ] is said to be @xmath2-uniformly distributed sequence ( abbreviated @xmath2-u.d.s . )
if for each @xmath83 with @xmath84 we have @xmath85)}{n}=d - c.\eqno ( 3.1)\ ] ] we denote by @xmath11 the set of all real valued sequences from @xmath0^{\infty}$ ] which are @xmath2-u.d.s .
it is well known that @xmath86 for each irrational number @xmath87 , where @xmath88 denotes the fractional part of the real number(cf .
@xcite , exercise 1.12 , p. 16 ) .
a sequence of real numbers @xmath89 is said to be uniformly distributed module @xmath5 if the sequence it s fractional parts @xmath90 is @xmath2-u.d.s .
it is obvious that @xmath91 is uniformly distributed module @xmath5 if and only if @xmath92 is @xmath2-u.d.s .
a sequence of real numbers @xmath82^{\infty}$ ] is said to be @xmath4-uniformly distributed sequence ( abbreviated @xmath4-u.d.s . )
if for each @xmath83 with @xmath84 we have @xmath85\cap c_t)}{n}=d - c.\eqno ( 3.2)\ ] ] we denote by @xmath93 the set of all real valued sequences from @xmath0^{\infty}$ ] which are @xmath4-u.d.s . in order to construct @xmath4-u.d.s . for each @xmath47 $ ] , we need the following lemma .
( @xcite , theorem 1.2 , p. 3 ) if the sequence @xmath94 is u.d .
mod 1 , and if @xmath95 is a sequence with the property @xmath96 for some real constant @xmath87 , then @xmath95 is u.d .
for each @xmath47 $ ] , there exists @xmath4-u.d.s .
let consider a sequence @xmath97 which is @xmath2-u.d.s .
for each @xmath98 , we choose such an element @xmath99 from the set @xmath100 that @xmath101 .
this we can do because @xmath102 is everywhere dense in @xmath103 .
now it is obvious that @xmath104 . by lemma 3.6
we deduce that @xmath95 is @xmath2-u.d.s .
let us show that @xmath95 is @xmath4-u.d.s .
indeed , since @xmath105 for each @xmath76 and @xmath95 is @xmath2-u.d.s .
, for each @xmath83 with @xmath84 we have @xmath106\cap c_t)}{n}=\ ] ] @xmath106)}{n}=d - c.\eqno ( 3.3)\ ] ] for each @xmath47 $ ] , @xmath4-u.d.s .
is @xmath2-u.d.s .. let @xmath107 be @xmath4-u.d.s .
on the one hand , for each @xmath83 with @xmath84 we have @xmath108)}{n}\ge\ ] ] @xmath85\cap c_t)}{n}=d - c.\eqno ( 3.4)\ ] ] since @xmath107 is @xmath4-u.d.s .
, we have @xmath109 \cap c_t)}{n}=1.\eqno ( 3.5)\ ] ] it is obvious that @xmath109)}{n}=1.\eqno ( 3.6)\ ] ] the last two conditions implies that @xmath110\setminus c_t ) ) } { n}=\ ] ] @xmath109)}{n}-\ ] ] @xmath111 the last relation implies that for each @xmath83 with @xmath84 @xmath85 \cap ( [ 0,1]\setminus c_t))}{n } \le\ ] ] @xmath110\setminus c_t))}{n}=0.\eqno ( 3.8)\ ] ] finally , for each @xmath83 with @xmath84 we get @xmath112)}{n}\le\ ] ] @xmath85\cap c_t)}{n } + \ ] ] @xmath85\cap ( [ 0,1 ] \setminus c_t))}{n}=\ ] ] @xmath113 this ends the proof of theorem .
note that the converse to the result of theorem 3.8 is not valid .
indeed , for fixed @xmath47 $ ] , let @xmath95 be @xmath4-u.d.s . which comes from theorem 3.7 . by theorem 3.8
, @xmath95 is @xmath2-u.d.s .
let us show that @xmath95 is not @xmath114-u.d.s . for each @xmath115 \setminus \{t\}$ ] .
indeed , since @xmath105 for each @xmath76 , we deduce that @xmath116 for each @xmath115 \setminus \{t\}$ ] .
the latter relation implies that for each @xmath115 \setminus \{t\}$ ] and for each @xmath83 with @xmath84 we have @xmath106\cap c_s)}{n}=0 < d - c.\eqno ( 3.10)\ ] ] for each @xmath2-u.d.s .
@xmath95 there exists a countable subset @xmath117 $ ] such that @xmath95 is not @xmath4-u.d.s for each @xmath47 \setminus t$ ] . indeed , since @xmath118\}$ ] is the partition of the @xmath0 $ ] , for each @xmath76 there exists a unique @xmath119 $ ] such that @xmath120 .
now we can put @xmath121 .
there exists @xmath2-u.d.s which is not @xmath4-u.d.s . for each @xmath47 $ ] .
let consider a sequence @xmath97 which is @xmath2-u.d.s .
since @xmath118\}$ ] is the partition of the @xmath0 $ ] , for each @xmath76 there exists a unique @xmath119 $ ] such that @xmath120 .
now we can put @xmath121 .
let @xmath122 be a countable subset of the set @xmath0\setminus t$ ] .
for each @xmath98 , we choose such element @xmath99 from the set @xmath123 that @xmath101 .
this we can do because @xmath102 is everywhere dense in @xmath103 for each @xmath47 $ ] .
now it is obvious that @xmath104 . by lemma 3.6
we deduce that @xmath95 is @xmath2-u.d.s .
let us show that @xmath95 is not @xmath4-u.d.s . for each @xmath47 $ ] .
this follows from the fact that @xmath124 for each @xmath47 $ ] . by this reason for each
@xmath47 $ ] and for each @xmath83 with @xmath84 we have @xmath106\cap c_t)}{n}\le\ ] ] @xmath125 @xmath126 for each different @xmath127 $ ] .
assume the contrary and let @xmath128 .
on the one hand , for each @xmath83 with @xmath84 we have @xmath85 \cap c_i)}{n}= d - c.\eqno ( 3.12)\ ] ] on the other hand , for same @xmath83 we have @xmath85 \cap c_j)}{n}= d - c .
\eqno ( 3.13)\ ] ] by theorem 3.8 we know that @xmath107 is @xmath2-u.d.s .
which implies that for same @xmath83 we have @xmath85)}{n}= d - c.\eqno ( 3.14)\ ] ] but @xmath129 is not possible because @xmath130 which implies @xmath131)}{n } \ge\ ] ] @xmath85 \cap c_i)}{n } + \ ] ] @xmath85 \cap c_j)}{n}=\ ] ] @xmath132 we get the contradiction and theorem is proved .
we have the following version of hlawka s theorem(cf .
@xcite ) for @xmath4-uniformly distributed sequences . for @xmath47 $ ] , we have @xmath133 .
let @xmath134 be a countable subclass of @xmath135 , \lambda_t)$ ] which defines a @xmath4-uniform distribution on @xmath0 $ ]
. of elements of @xmath135 , \lambda_t)$ ] defines a @xmath4-uniform distribution on @xmath0 $ ] , if for each @xmath7^{\infty}$ ] the validity of the condition @xmath136 } f_k(x)d \lambda_t(x)$ ] for @xmath76 implies that @xmath94 is @xmath4-u.d.s .
indicator functions of sets @xmath137 \cap c_t$ ] with rational @xmath83 is an example of such a family . ] for @xmath138 , we set @xmath139^{\infty } ~\ & ~\lim_{n \to \infty}\frac{1}{n}\sum_{n=1}^nf_k(x_n)=\int_{[0,1 ] } f_k(x)d \lambda_t x\}.\ ] ] by lemma 3.1 we know that @xmath140 for @xmath141 which implies @xmath142 .
hence @xmath143^{\infty } ~\ & ~(\forall k)(k \in { \bf
n } \rightarrow \lim_{n \to \infty}\frac{1}{n}\sum_{n=1}^n f_k(x_n)=\ ] ] @xmath144 } f_k(x)d \lambda_t(x))\})=1.\ ] ] the latter relation means that @xmath3-almost every elements of @xmath0^{\infty}$ ] is @xmath4-u.d.s .
, equivalently , @xmath133 .
we have the following analogue of h.weyl theorem ( cf .
@xcite ) for @xmath4-uniformly distributed sequences . for @xmath47 $ ]
, we put @xmath145=\{\tilde{h}(x)= h(x ) \times \chi_{c_t}(x ) : h \in c[0,1]\}$ ] .
then the sequence @xmath94 is @xmath4-u.d.s . if and only if the following condition @xmath146}\tilde{h}(x)d\lambda_t(x)\eqno ( 3.16)\ ] ] holds for each @xmath147 $ ] .
let @xmath94 be @xmath4-u.d.s . and
let @xmath148 be a spatial step function on @xmath0 $ ] , where @xmath149 then it follows from ( 3.2 ) that for every such @xmath8 equation ( 3.16 ) holds .
we assume now that @xmath150 $ ] .
given any @xmath151 , there exist , by the definition of the riemann integral , two step functions , @xmath152 and @xmath153 say , such that @xmath154 for all @xmath155 $ ] and @xmath144}(f_2(x ) -f_1(x ) ) d\lambda(x ) < \epsilon.\ ] ] then it is obvious that @xmath156 for all @xmath155 $ ] and @xmath144}(f_2(x)\chi_{c_t}(x ) -f_1(x)\chi_{c_t}(x ) ) d\lambda_t(x ) = \int_{[0,1]}(f_2(x ) -f_1(x ) ) d\lambda(x ) < \epsilon.\ ] ] then we have the following chain of inequalities : @xmath144}\tilde{f}(x)d\lambda_t(x)-\epsilon \le \int_{[0,1]}f_1(x)\chi_{c_t}(x)d\lambda_t(x)=\ ] ] @xmath157 @xmath158 @xmath159 @xmath160 @xmath161 @xmath162 @xmath144}f_2(x)\chi_{c_t}(x)d\lambda_t(x)\le\ ] ] @xmath144}\tilde{f}(x)d\lambda_t(x)+\epsilon.\eqno ( 3.18)\ ] ] so that in the case of a function @xmath163 the relation ( 3.16 ) holds .
conversely , let a sequence @xmath94 be given , and suppose that ( 3.16 ) holds for every @xmath150 $ ] .
let @xmath164 be an arbitrary subinterval of @xmath0 $ ] .
given any @xmath151 , there exist two continuous functions , @xmath165 and @xmath166 say , such that @xmath167 for @xmath155 $ ] and at the same time @xmath168}(g_2(x ) - g_l(x ) ) d\lambda x < \epsilon $ ] .
note that at the same time we have @xmath169 for @xmath155 $ ] and @xmath168}(\tilde{g}_2(x ) - \tilde{g}_l(x ) ) d\lambda_t x < \epsilon$ ] .
then we get @xmath170}g_2(x ) d \lambda x - \epsilon <
\int_{[0,1]}g_1(x ) d \lambda x = \ ] ] @xmath144}\tilde{g}_1(x ) d \lambda_t x = \lim_{n \to \infty}\frac{\sum_{n=1}^n\tilde{g}_1(x_n)}{n}\le\ ] ] @xmath171 @xmath172 @xmath173 @xmath144}\tilde{g}_2(x ) d \lambda_t x
= \int_{[0,1]}g_2(x ) d \lambda x\ ] ] @xmath174}g_1(x ) d \lambda x + \epsilon \le\ ] ] @xmath144}\chi_{[a , b[}(x ) d \lambda x + \epsilon= b - a + \epsilon.\eqno ( 3.19)\ ] ] since @xmath175 is arbitrarily small , we have ( 3.2 ) .
by vitali s celebrate theorem about existence of the linear lebesgue non - measurable subset has been shown that the domain of the lebesgue measure in @xmath20 differs from the power set of the real axis @xmath20 . in this context
the following question was naturally appeared : _ how far can we extend lebesgue measure and what properties can such an extension preserve ?
_ in 1935 e. marczewski , applied sierpiski construction of an almost invariant set @xmath176 , obtained a proper invariant extension of the lebesgue measure in which the extended @xmath22-algebra had contained new sets of positive finite measure . in connection with this result ,
waclaw sierpiniski in 1936 posed the following * problem ( waclaw sierpiiski ) * _ let @xmath177 denotes the group of all isometrical transformations of the @xmath178 .
does there exist any maximal @xmath177 -invariant measure ? _
the first result in this direction was obtained by andrzej hulanicki @xcite as follows : * proposition * ( andrzej hulanicki ( 1962 ) ) _ if the continuum @xmath179 is not real valued measurable cardinal then there does not exist any maximal invariant extension of the lebesgue measure .
_ this result was also obtained independently by s. s. pkhakadze @xcite using similar methods . in 1977
a. b. kharazishvili got the same answer in the one - dimensional case without any set - theoretical assumption ( see @xcite ) . finally , in 1982 krzysztof ciesielski and andrzej pelc generalized kharazishvili s result to all @xmath19-dimensional euclidean spaces ( see @xcite ) .
following solovay @xcite , if the system of axioms @xmath180 there exists inaccessible cardinal is consistent then the systems of axioms
@xmath181 every set of reals is lebesgue measurable is also consistent .
this result implies that the answer to waclaw sierpiniski s problem is affirmative .
taking solovay s result on the one hand , and krzysztof ciesielski and andrzej pelc ( or andrzej hulanicki or pkhakadze ) result on the other hand , we deduce that the waclaw sierpiniski s question is not solvable within the theory @xmath182 now days there exists a reach methodology for a construction of invariant extensions of the lebesgue measure in @xmath178 as well the haar measure in a locally compact hausdorff topological group .
let us briefly consider main of them .
* method i. * ( jankowka - wiatr ) following @xcite , the first idea of extending the lebesgue measure in @xmath178 to a larger @xmath22-algebra in such a what that it remains invariant under translations belongs to jankowka - wiatr who in 1928 observed that one can add new sets to the @xmath22-ideal of sets of lebesgue measure zero and still preserve the invariance of the extended measure .
this method can be described as follows : _ let @xmath23 be a shift - invariant @xmath22-ideal in the @xmath19-dimensional euclidean space @xmath178 such that @xmath184 where @xmath185 denotes the inner measure defined by @xmath19-dimensional lebesgue measure @xmath63 . then the functional @xmath186 defined by @xmath187 where @xmath28 is a borel subset of @xmath178 and @xmath29 and @xmath30are elements of the @xmath22-ideal @xmath23 , is an @xmath177-invariant extension of the lebesgue measure measure @xmath63 . _ * method ii . *
( e. szpilrajn(e .
marczewski ) ) by using sierpiski s decomposition @xmath188 of the @xmath189 , e. szpilrajn noted that the following two conditions \(i ) @xmath190 for each @xmath191 ; \(ii ) @xmath192 for each closed set @xmath193 whith @xmath194 .
holds true .
further , he constructed a proper shift - invariant extension @xmath186 of the lebesgue measure @xmath63 in @xmath189 as follows @xmath195 for @xmath196 .
* method iii . *
( oxtoby and kakutani ) some methods of combinatorial set theory have lately been successfully used in measure extension problem . among them , special mention should be made of the method of constructing a maximal ( in the sense of cardinality ) family of independent families of sets in arbitrary infinite base spaces .
the question of the existence of a maximal ( in the sense of cardinality ) @xmath197-independent of subsets of the set @xmath16 is @xmath197-independent if the condition @xmath198 holds , where @xmath199 if in addition , this condition holds true for each @xmath200 with @xmath201 , then @xmath202 is called strict @xmath197-independent.]family of subsets of an uncountable set @xmath16 was considered by a. tarski .
he proved that this cardinality is equal to @xmath203 .
this result found an interesting application in general topology .
for example , it was proved that in an arbitrary infinite space @xmath16 the cardinality of the class of all ultrafilters is equal to @xmath204 ( see , e.g. , @xcite ) .
the combinatorial question of the existence of a maximal ( in the sense of cardinality ) strict @xmath197-independent family of subsets of a set @xmath16 with cardinality of the continuum also was investigated and was proved that this cardinality is equal to @xmath205 .
this combinatorial result found an interesting application in the lebesgue measure theory .
for example , kakutani and oxtoby @xcite firstly constructed a family @xmath206 of almost invariant subsets of the circle in such a way that @xmath207 has outer measure @xmath5 for an arbitrary sequence @xmath208 of sets from @xmath206 and arbitrary sequence @xmath209 .
the putting @xmath210 for @xmath176 in @xmath206 they obtained an extension of the lebesgue measure on the circle to an invariant measure @xmath186 such that @xmath211 has the hilbert space dimensional equal to @xmath205 . using the same combinatorial result , a.b .
kharazishvili constructed a maximal ( in the sense of cardinality ) family of orthogonal elementary @xmath177-invariant extensions of the lebesgue measure ( see @xcite ) .
the combinatorial question of the existence of a maximal ( in the sense of cardinality ) strict @xmath197-independent family of subsets of a set @xmath16 with @xmath212 was investigated in @xcite and it was shown that this cardinality is equal to @xmath203 . using this result ,
g.pantsulaia @xcite extended kakutani and oxtoby @xcite method for a construction of a maximal ( in the sense of cardinality ) family of orthogonal elementary @xmath213-invariant extensions of the haar measure defined in a locally compact @xmath22-compact topological group with @xmath214 .
* method iv . *
( kodaira and kakutani method ) kodaira and kakutani @xcite invented the following method of extended the lebesgue measure on the circle to an invariant measure as follows : let produce a _ character _
@xmath215 of the circle , i.e. a homomorphism @xmath216 in such a way that the outer lebesgue measure of its graph @xmath217 is equal to @xmath5 in @xmath218 .
then the extended @xmath22-algebra @xmath219 consists of sets @xmath220 , where @xmath221 is lebesgue measurable set in @xmath218 and the extended measure @xmath186 is @xmath222 note that the discontinuous character @xmath215 becomes @xmath219-measurable .
it has been noticed later in @xcite that one can produce @xmath205 such characters so that they all become measurable and @xmath223 is of hilbert space dimension @xmath205 .
this method have been modified for @xmath19-dimensional euclidean space in @xcite(received 7 .
november 1993 ) for a construction of the invariant extension @xmath26 of the @xmath19-dimensional lebesgue measure such that there exists a @xmath26-measurable set with only one density point .
this result answered positively to a certain question stated by a.b .
kharazishvili ( cf .
@xcite , problem 9 , p. 200 ) . knowing this result ,
kharazishvili considered similar but originally modified method and extended previous result in @xcite ( received 15 .
march 1994 ) as follows : _ there exists an invariant extension @xmath26 of the classical lebesgue measure such that @xmath26 has the uniqueness property and there exists a @xmath26-measurable set with only one density point._. * method @xmath224 * . more lately , kodaira and kakutani method have been modified for an uncountable locally compact @xmath22-compact topological group @xmath213 with @xmath225 in @xcite as follows : let @xmath16 be a set with @xmath226 and let @xmath26 be a probability measure in @xmath16 such that each @xmath227 for which @xmath228 .
let produce a _ function _ @xmath229 in such a way that the following two conditions @xmath230 is a closed subset of the @xmath213 with @xmath231 @xmath232 holds true .
then the extended @xmath22-algebra @xmath219 consists of sets @xmath233 , where @xmath234 .
then the extended measure @xmath235 is defined by @xmath236 note that @xmath235 is a non - elementary invariant extension of the measure @xmath63 iff the measure @xmath26 is diffused .
it has been noticed that one can produce @xmath237 such functions so that they all become measurable and @xmath238 is of hilbert space dimension @xmath237 . note that when @xmath239 , @xmath26 is a normalized counting measure in @xmath16 and @xmath240 is defined by @xmath241 , then method @xmath224 gives marczewski method .
when @xmath242 , @xmath243 and @xmath244 is a character , then method @xmath224 gives kodaira and kakutani method .
now let us discuss whether method @xmath224 gives oxtoby and kakutani method .
in this context we need some auxiliary facts .
( @xcite , * theorem 11.1 * , p. 158 ) if an infinite set @xmath16 satisfies the condition @xmath245 then there exists a maximal ( in the sense of cardinality ) strictly @xmath197-independent family @xmath246 of subsets of the space @xmath16 , such that @xmath247 ( @xcite , * lemma 11.2 * , p. 163 ) let @xmath213 be an arbitrary locally compact @xmath22-compact topological group , @xmath2 be the haar measure defined on the group @xmath213 and let @xmath87 be an arbitrary cardinal number such that : @xmath248 then there exists a family @xmath202 of subsets of the set @xmath213 such that : @xmath249 @xmath250 @xmath251 is a closed subset of the space @xmath213 with @xmath252 @xmath253 @xmath254 ( @xcite , * lemma 11.5 * , p. 169 ) let @xmath16 be an uncountable base space with @xmath255 .
then there exists a non - atomic probability measure @xmath256 such that the following conditions hold : @xmath257 @xmath258 the topological weight @xmath259 of the metric space @xmath260 associated with measure @xmath261 is maximal , in particular , is equal to @xmath203 .
now put @xmath262 and @xmath263 . for @xmath264
, we put @xmath265 , where @xmath266 is a unique index for which @xmath267(cf .
lemma 4.2 for @xmath268 ) .
we set @xmath269 for @xmath270 .
let @xmath271 comes from 4.3 .
now if we consider the invariant extension @xmath272 of the measure @xmath63 , we observe that @xmath273 for each @xmath274 . since @xmath206 is the family of strictly @xmath197-independent almost invariant subsets of @xmath275
, we claim that method @xmath224 gives just above described oxtoby and kakutani method .
* method v. * ( kharazishvili ) .
this approach , as usual , can be used for uncountable commutative groups and is based on purely algebraic properties those groups , which are not assumed to be endowed with any topology but only are equipped with a nonzero @xmath22-finite invariant measures . here
essentially is used kulikov s well known theorem about covering of any commutative group by increasing ( in the sense of inclusion ) countable sequence of subgroups of @xmath275 which are direct sum of cyclic groups ( finite or infinite ) ( see , for example @xcite , @xcite ) . * definition 4.4(krarazishvili )
* let @xmath16 be a base space , @xmath275 be a group of transformations of @xmath16 and let @xmath28 be a subset of the space @xmath16 .
@xmath28 is called a @xmath275-absolutely negligible set if for any @xmath275-invariant @xmath22-finite measure @xmath26 , there exists its @xmath276invariant extension @xmath277 such that @xmath278 and @xmath279 . a geometrical characterization of absolutely negligible subsets , due to a.b .
kharazishvili , is presented in the next proposition .
* theorem 4.5 * _ let @xmath16 be a base space , @xmath275 be a group of transformations of @xmath16 containing some uncountable subgroup acting freely in @xmath16 , and @xmath28 be an arbitrary subset of the space @xmath16 .
then the following two conditions are equivalent : _
1 ) x is a @xmath275-absolutely negligible subset of the space @xmath16 ; 2 ) for an arbitrary countable @xmath275-configuration of @xmath16 is called a countable @xmath275-configuration of @xmath28 if there is a countable family @xmath280 of elements of @xmath275 such that @xmath281 . ] @xmath282 of the set @xmath28 , there exists a countable sequence @xmath283 of elements of @xmath275 @xmath284 it is of interest that the class of all countable @xmath275-configurations of the fixed @xmath275-absolutely negligible subset constitutes a @xmath275-invariant @xmath22-ideal such that the inner measure of each element of this class is zero with respect to any @xmath22-finite @xmath275-invariant measure in @xmath16 .
hence , by using the natural modification of the method i one can obtain @xmath275 -invariant extension of an arbitrary @xmath22-finite @xmath275-invariant measure in @xmath16 . in 1977
a. b. kharazishvili constructed the partition of the real axis @xmath20 in to the countable family of @xmath285 -absolutely negligible sets and got the negative answer to the question of waclaw sierpiiski in the one - dimensional case without any set - theoretical assumption ( see @xcite ) . finally , in 1982
krzysztof ciesielski and andrzej pelc generalized kharazishvili s result to all @xmath19-dimensional euclidean spaces , more precisely , they constructed the partition of the euclidean space @xmath178 in to the countable family of @xmath177 -absolutely negligible sets and got the negative answer to the question of waclaw sierpiiski in the @xmath19-dimensional case without any set - theoretical assumption ( see @xcite ) . by using the method of absolutely negligible sets elaborated by a.kharazishvili @xcite ,
p. zakrzewski @xcite answered positively to a question of ciesielski @xcite asking _ whether an isometrically invariant @xmath22-finite countably additive measure on @xmath178 admits a strong countably additive isometrically invariant extension .
_ it is obvious that this question is generalization of the above mentioned waclaw sierpiiski problem .
let @xmath275 be a compact hausdorff topological group and @xmath2 be a haar measure in @xmath213 .
by @xmath286 we mean the set of all bounded real - valued borel - measurable functions on @xmath275 . under the norm @xmath287 for @xmath288 , the set @xmath286 forms a banach space , and even a banach algebra if algebraic operations for functions are defined in the usual way .
the subset @xmath289 of @xmath286 consisting of all real - valued continuous functions on @xmath275 is then a banach subalgebra of @xmath286 .
note that the theory of uniform distribution is well developed in compact hausdorff topological groups ( see , for example @xcite , chapter 4 ) as well the theory of invariant extensions of haar measures in the same groups(see section 4 ) . here
naturally arise a question asking _ whether can be introduced the concept of uniform distribution for invariant extensions of the haar measure in compact hausdorff topological groups .
_ we wait that by using similar manner used in section 3 and the methodology briefly described in section 4 , one can resolve this question . | the concept of uniform distribution in @xmath0 $ ] is extended for a certain strictly separated maximal ( in the sense of cardinality ) family @xmath1}$ ] of invariant extensions of the linear lebesgue measure @xmath2 in @xmath0 $ ] , and it is shown that the @xmath3 measure of the set of all @xmath4-uniformly distributed sequences is equal to @xmath5 , where @xmath3 denotes the infinite power of the measure @xmath4 .
this is an analogy of hlawka s ( 1956 ) theorem for @xmath4-uniformly distributed sequences .
an analogy of weyl s ( 1916 ) theorem is obtained in the similar manner . |
solitons , as solutions of non - linear wave equations have , by now , been studied extensively for many years @xcite .
however , they are still producing new and unexpected phenomena . such is , for instance , their behaviour when one sends a soliton towards a potential obstruction as was studied in @xcite , @xcite . the results presented in these papers were obtained for solitons in ( 2 + 1 ) dimensions and for the sine - gordon model in ( 1 + 1 ) dimensions . in a recent study
@xcite we looked at a similar scattering of solitons in two ( 1 + 1 ) dimensional @xmath5 models . in these models
we have inserted potential obstructions in two different ways , in each case modifying the lagrangian of the model in a region far away from the soliton . in the first model
this modification was introduced via the coupling of the potential ( which was made to be position dependent ) , while in the second model this was achieved via the modification of the minkowski space - time metric @xcite . in each model , like in @xcite and @xcite , the topological solitons were found to scatter on the barrier in a very elastic way : they can either overcome the barrier and get transmitted or get reflected from the barrier with almost no loss of energy .
however , the scattering from a hole was found to be inelastic and , at times , produced an almost non - classical behaviour as solitons with energies below a critical value for the transmission could either be trapped in the hole or get reflected ! this reflection resembles a little the quantum reflection and so ,
perhaps , some classical phenomena based on solitons could be confused with their quantum behaviour . in view of this
we have decided to look at other models ; to see whether the observed phenomena are ` universal ' . a good class of such models is provided by the models of bazeia et al @xcite .
not only they depend on a parameter ( which takes integer values ) ; they also reduce to the well known sine - gordon model for a specific value of this parameter . hence in this paper
we look at the scattering of solitons in these models on both potential holes and potential barriers . like in the original work on the sine - gordon model @xcite the obstructions are introduced via the modification of the coupling constant in the models _
i.e. _ by making it space dependent . in the next section
we introduce the models of bazeia et al @xcite and discuss some of their properties .
we present the relevant lagrangians ( for both classes of generalizations of the sine - gordon model ) and discuss their soliton ( kink - like ) solutions .
the following two sections discuss our results on the scattering of these solitons ( from each class ) on both types of obstructions .
we finish the paper with a short section presenting our conclusions .
we consider , in full generality , a single real scalar field in ( 1 + 1 ) dimensions @xmath6 where the potential @xmath7 is taken in the form @xmath8 , in which @xmath9 . here
@xmath10 is an extra potential parameter which has been inserted into the potential @xmath7 to take into account the effects of obstructions , holes and barriers , and so is nonzero only in a certain region of space . in our case
, we put the obstruction around the origin ( _ i.e. _ @xmath11 ) so we take @xmath12 the equation of motion is , of course , @xmath13 where @xmath14 is the derivative of @xmath15 with respect to the argument .
the models of bazeia et al @xcite correspond to choosing ( ` type 1 ' models ) @xmath16 of the form : @xmath17 when @xmath1 this model reduces to the sine - gordon one .
the second class ( ` type ii models ' ) correspond to taking @xmath16 in the form @xmath18 in this case the sine - gordon model corresponds to @xmath2 .
both classes of models are topological as in each case we can write for the static fields : @xmath19 the last term is a total divergence and so we see that the equation for the static solitons ( bogomolnyi equation ) becomes @xmath20 the solutions of ( [ bogo ] ) are easy to find in the case when we have no obstruction ( _ i.e. _ when @xmath21 ) .
they are given by ( for the first class of models ) : @xmath22^\frac{1}{\,n } , \label{onee}\ ] ] where , due to the lorentz invariance of the full lagrangian ( [ lag ] ) , we have inserted the time dependence by performing the relativistic boost . in ( [ onee ] )
@xmath23 is the usual relativistic factor , _ ie _ @xmath24 .
the expression ( [ onee ] ) is , in fact , a solution of ( [ bogo ] ) when @xmath21 .
it is also an approximate solution of ( [ bogo ] ) when @xmath25 for @xmath26 far away from the region in which @xmath27 . the soliton solutions ( [ onee ] ) satisfy the kink boundary conditions @xmath28 .
for the second class of models the soliton solutions ( again with the same comments about the @xmath29 dependence of @xmath30 ) are given by : @xmath31^\frac{1}{\,n},\ ] ] and they satisfy the following boundary conditions : @xmath32 . in figures 1 and 2 we present plots of the static soliton field configurations for the type i and type ii models , respectively , for the first 6 models in each class , _
i.e. _ for @xmath33 . in the next section
we look at the scattering properties of these solitonic solutions in the first class of models .
= 1, ... ,6.,width=302 ] = 1, ... ,6.,width=302 ]
before we consider the scattering properties of the solitonic solutions let us observe that the mass ( rest energy ) of a static soliton is given by @xmath34\,dx.\ ] ] the energy of a soliton moving with velocity @xmath35 is then given by @xmath36 to calculate @xmath37 we note that it can be rewritten as @xmath38 where @xmath39 is @xmath40 . to perform the calculation of ( [ mass ] ) it is convenient to change variables from @xmath29 to @xmath41 .
then ( [ mass ] ) becomes proportional to @xmath42 _ i.e. _ @xmath43 this integral defining @xmath42 can be performed explicitly for each value of @xmath0 .
we find that @xmath44 and all others satisfy the recurrence relation @xmath45\frac{1}{n+2}.\ ] ] in figure 3 we plot the rest mass energies for @xmath33 .
the figure shows that solitons becomes less massive as @xmath0 increases .
the figure also shows that the @xmath2 the soliton is very massive .
, width=302 ] next we considered the scattering properties of these solitons . to do this
we put our obstructions close to @xmath11 ( in fact between -5 and 5 ) and initially placed solitons far away from @xmath11 , namely around @xmath46 sending them with some velocity @xmath35 towards the obstruction . in our numerical work
we used the square well potential barriers and holes of width 10 .
the width was chosen carefully so that the solitons can fit into the potential hole and have enough space to move inside the hole or on the top of the barrier .
the simulations were performed using the 4th order runge - kutta method of simulating the time evolution .
we used 1201 points with the lattice spacing of @xmath47 .
hence , the lattice extended from -60 to 60 in the @xmath29-direction .
the time step was chosen to be @xmath48 . in our work
we used the absorbing boundary conditions .
our simulations have shown that , like in the previous work on the sine - gordon @xcite and @xmath49 @xcite models , for any @xmath0 , there is a critical velocity above which the solitons are transmitted by the hole and below which they are either trapped or reflected .
see also @xcite and @xcite which present earlier work on this subject with an interesting explanation of the reflection property of the hole in terms of the interference of the soliton field with the radiation waves generated by the scattering .
hence we believe that the observed phenomenon is a generic behaviour of topological solitons as they encounter a potential hole .
the value of the critical velocity depends on the model in question .
the scattering is inelastic with the soliton emitting radiation both in the hole and / if when it has left the hole .
the amount of the emitted radiation is expected to be related to the integrability , or not , of the model in question .
hence to study this further we have decided to look in detail on the emission of radiation in the models of bazeia et al .
thus we have examined in detail the scattering properties of solitons for @xmath50 .
first we looked at the values of the critical velocity for the hole of depth = -0.50 ( _ i.e. _ @xmath51 ) .
figure 4 gives the plot of these values and we note that the critical velocity is the lowest for @xmath4 . in order to have a better understanding to our results on the critical velocities , we have looked into the energy of solitons after the scattering when they are already far away from the hole .
the total energy of the solitons after the scattering is given by @xmath52 we have estimated the ratios of the radiation to the vibrational energies .
table 1 presents the results of our calculations . [ cols= " < , < " , ] clearly , both sets of values of critical velocities are in a very good agreement . at the same time , the facts that the approximate value is lower , but only marginally so , and the value of this difference , shows that the radiation effects are very small though , strictly speaking , nonzero .
we have also looked at the scattering of solitons on holes and barriers in the second class of models of bazeia et al .
most of the behaviour was very similar to what was seen in the first class of models ( _ i.e. _ the existence of critical velocities , transmission and reflections on barriers etc ) .
but there were also some small differences .
hence here we restrict our discussion to the description of these differences .
the main reason for these differences resides in the form of the soliton field itself , and the fact that , as shown in figure 2 , the fields are very asymmetric with respect to their behaviour as @xmath53 . clearly as @xmath0 increases
the fields go to different , and decreasing values as @xmath54 .
similarly , as @xmath0 increases , they go to 0 much more slowly .
hence the energy densities of the solitons are more spread out , as shown in figure 11 .
= 4,5,6.,width=302 ] moreover , like for the first class of models , the rest masses of solitons decrease with @xmath0 , but this time this decrease is faster . in fig 12 .
we present the rest mass energies of solitons as a function of @xmath0 .
clearly , we see a big decrease as @xmath0 increases . as we will show below the tail of solitons in class ii models , except for @xmath2 when the model reduces to the sine - gordon one , affects the behaviour of solitons when they scatter on a potential hole .
the effect increases with @xmath0 as for larger @xmath0 solitons are more spread out and also are less massive . in figure 11 we plot the energy densities of the soliton solutions of type ii models .
we only plot them for @xmath55 and @xmath56 as the energy densities for lower values on @xmath0 are much larger so that plotting them all together would make the plots of larger @xmath0 almost invisible . a note on the normalisation .
as we said before the @xmath1 type i model and @xmath2 type ii one are the same - in fact they both correspond to the sine - gordon model . however , our values of their enegies are different ( in the type i case the rest energy is 0.5 , while in the type ii it is 2.0 ) .
the difference comes from the different normalisation of both models ( @xmath57 ) .
= 1, ... ,6.,width=302 ] figures 13 and 14 present the plots of the critical velocities when the solitons are scattered from a hole of -0.5 depth and a barrier of 0.4 height , respectively .
the tail , as we have observed in our numerical simulations , is the main part of the solitons that is the most affected when the solitons are scattered on a hole .
more precisely , our simulations have shown that the front part of the soliton is less affected by the hole while the tail , which is its rear part , is more affected by it .
this is due to the asymmetry of the soliton field configuration which generates this asymmetry of the energy density of the soliton .
thus , for example , the energy radiated by the soliton comes mainly from the tail .
when the velocity of the incoming soliton is close to its critical value the tail get badly deformed and the energy radiated from the tail is large . below the critical velocity , _
i.e. _ when the soliton gets trapped inside the hole we have sometimes observed that the peak of the radiated energy is greater than the maximum of energy density of the soliton itself . in such cases
one might even observe a false quantum - like reflection .
this false reflection can be observed numerically unless one is careful and looks at the fields in some detail . while the soliton is inside the hole the magnitude of the back radiation may suggest its reflection
. only careful analysis of such a case would clarify the situation . in one case to be absolutely certain what has happened we had to make the hole very wide to be certain that the soliton remained trapped in it .
this was the case , for example , when we looked at the @xmath58 model and the soliton was sent with velocity @xmath59 towards a hole of depth -0.5 .
this soliton got trapped inside the hole and radiated away a very large amount of radiation .
figure 13 shows the energy density seen in this case when the radiated energy which has travelled backwards is higher than the main energy of soliton which is trapped inside the hole .
one can also see from this figure that the front part of the soliton has not shown any significan deformation while the tail part of the energy density has been deformed greatly and has given off almost all the radiated energy .
= 6.,width=170 ] we have also observed that in the cases when the velocity is below its critical values the front part of the soliton including its centre of mass can exit the hole while the tail still remains inside it .
then , because the tail has remained in the hole , the soliton gets pulled back into the hole and so remains trapped inside it .
we have considered two classes of topological soliton models , presented by bazeia et al @xcite which are generalisations of the sine - gordon model .
both classes depend on an integer parameter @xmath0 which , for the type i models reduces to the sine - gordon model when @xmath1 , while in the type ii case the sine - gordon case corresponds to @xmath0=1 .
both classes of models have shown a behaviour which is similar to what was observed in the sine - gordon case in some recent research work , namely that the scattering by a hole is inelastic while the scattering by a barrier is nearly elastic . during the scattering on barriers
the solitons behave very much like point - like particles .
their initial kinetic energy is converted into the potential energy needed to overcome the barrier and there is very little energy left to excite the internal degrees of freedom of the soliton .
this has been confirmed by our estimates of the critical velocities for the transmission over the barrier which is based on this assumption .
we have shown this to be true in all models ( _ i.e. _ for all values of @xmath0 ) .
the scattering on the holes is different . when a soliton enters a hole it gains an extra energy which is then , in part , converted into the energy of internal oscillations
this is again seen for all the models we have looked at .
why this extra energy is used this way is not completely clear at the moment .
we have tried to find an explanation in terms of equipartition of energy etc but have not managed to prove that this is really the case .
hence this explanation is still lacking and our results have shown that this is a general phenomenon as the excitation of the internal degrees of freedom takes place in the all models studied by us and is more or less similar in magnitude .
we have also seen that the energy of solitons after the scattering is distributed differently into the vibrational and radiation energies .
and so the soliton which converts most of its extra energy into the vibrational energy , rather than the radiation energy , is more able to be transmitted over the obstruction with the least velocity .
solitons of the type - ii models , except for the sine - gordon case ( @xmath2 ) , have asymmetrical field configurations and have exhibited what we have called the tail effect in their scattering on potential holes .
the perturbation due to the potential hole has an uneven effect on the body of solitons .
while it produces very little deformation or the radiation in the front part of the solitons , their tails are greatly deformed and most of the radiation is produced there too .
when the initial velocities of the solitons are close to their critical values , the radition energy , which is sent backwards , is very large and often more peaked than the energy density of the soliton itself .
this could easily lead to some misunderstanding of what has happened ; _
i.e. _ one may think that one has observed a quantum - like reflection while in reality this was only a trapping accompanied by some radiation sent backwards .
solitons , in some cases , behave like particles because of their localized structure . however , because of their extened structure they deviate from their particle behaviour in other cases .
the symmetrical or asymmetrical field configurations play a great role on how the perturbations affect the solitons and the amount of radiation that is produced . when a topological soliton ( corresponding to @xmath21 ) is placed inside a potential hole it has an extra energy .
this energy is then transfered into its vibrational energy but some of it is radiated off . moreover ,
when we looked at the energy of solitons scattered by a hole we have found it to be almost the same as the energy of the solitons before the scattering although the final solitons were moving with somewhat smaller velocities .
this implies that the scattering generated very little radiation ( _ i.e. _ also when the solitons were in the hole ) and that most of the extra energy was converted into the vibrational / excitation energy of the solitons and suprising little was sent off as radiation .
we have also observed that for @xmath4 type i model , the critical velocity is the least for all the cases we have looked at when considering the scattering of a soliton on a hole of -0.5 depth or on a barrier of 0.4 height .
piette , w.j .
zakrzewski and j. brand , _ j. phys .
_ * a 38 * , 10403 - 10412 ( 2005 ) b.m.a.g .
piette and w.j .
zakrzewski , _ j. phys . _ * a 40 * , 329 - 346 ( 2007 ) jassem h. al - alawi , w.j .
_ j. phys . _ * a 40 * , 11319 - 1131(2007 ) | we study scattering properties of topological solitons in two classes of models , which are the generalisations of the sine - gordon model and which have recently been proposed by bazeia et al .
these two classes of models depend on an integer parameter @xmath0 which , when @xmath1 ( for the first class ) and @xmath2 ( for the second class ) , reduce to the sine - gordon model .
we take the soliton solutions of these models ( generalisations of the ` kink ' solution of the sine - gordon model ) and consider their scattering on potential holes and barriers .
we present our results for @xmath3 .
we find that , like in the sine - gordon models , the scattering on the barrier is very elastic while the scattering on the hole is inelastic and can , at times , lead to a reflection .
we discuss the dependence of our results on @xmath0 and find that the critical velocity for the transmission through the hole is lowest for @xmath4 . |
the virgo cluster is an important stepping stone in the determination of cosmological distances and of the expansion parameter ( hubble constant ) of the universe ( e.g. tully 1988 , tammann 1991 ) .
an important objective of the hubble space telescope mission is the extension of the determination of cosmological distances by means of the period - luminosity relation of cepheid variables out to the distance of the virgo cluster of galaxies .
recently , the first hst cepheid distance measurement to the virgo galaxy m100 was reported ( freedman 1994 , mould 1995 , ferrarese 1996 ) .
the distance of the galaxy was determined to be @xmath3 mpc and a hubble constant of @xmath0 km s@xmath2 mpc@xmath2 was deduced from this result .
the most recent update on this results by ferrarese 1996 gives a smaller distance of @xmath4 mpc and a slightly higher implied hubble constant .
similarly in an earlier ground based study of the virgo spiral ngc4571 by pierce ( 1994 ) a cepheid distance of @xmath5 mpc was reported with a formally derived value of @xmath1 .
how the distance of m100 is related to the distance of the virgo cluster as a whole is a very crucial step in this derivation of the hubble constant .
since the mean expected peculiar motion of m100 is too large to yield a precise measure of @xmath6 directly , the cepheid distance measurement is used to infer the distance of the virgo cluster . to obtain a value for @xmath6 the recession velocity of the virgo cluster ( corrected for the `` virgo infall '' )
is either used directly or the recession velocity of the coma cluster is combined with the relative distance of the virgo cluster and the coma cluster in the paper by freedman ( 1994 ) to derives values of of @xmath7 and @xmath8 km s@xmath2 mpc@xmath2 , respectively .
an alternative and very promissing way to derive the `` cosmic recession velocity '' of the virgo cluster for this application is outlined by jerjen & tammann ( 1993 ) .
they use absolute distance measurements to 17 clusters to establish a `` machian frame '' which allows to separate the virgo recession and the local group peculair motion . with the velocity values given by jerjen &
tammann a lower value for @xmath6 is actually derived .
appart from this latter problem of separating the peculiar motion from the cosmic expansion in the local universe , the establishment of a proper value for the virgo distance is the most important , uncertain step in the derivation of @xmath6 as described above . in the following
we will therefore discuss if this determination can be improved . using a cepheid distance measurement to obtain the virgo distance one has to make the assumption that the observed spiral galaxy , e.g. m100 , has a certain line - of - sight distance from the center of the virgo cluster .
freedman ( 1994 ) in particular assume that m100 is located in the cloud of spiral galaxies centered on the virgo cluster with a line - of - sight extension of 6 mpc .
the lack of information on the precise location of m100 thus adds to the uncertainties of the determination of @xmath6 a value of 17.5% in the paper by freedman ( 1994 ) .
this assumption is made somewhat _
ad hoc_. spiral galaxies are not part of the central region of the virgo cluster but rather avoid the densest regions ( dressler 1984 , schindler 1995 ) .
mould ( 1995 ) for example show that a virgo distance of even as far as 20 mpc is still just consistent with the measured m100 cepheid distance for an approximately 90% confidence interval .
therefore the inability to locate precisely the galaxy with measured cepheid distances within the virgo cluster is a severe limitation to the determination of the distance to the virgo cluster . in the meantime also
other cepheid distances to virgo cluster spirals have been derived using the hubble space telescope .
the galaxies ngc 4496 , ngc 4536 , and ngc 4639 were observed by saha ( 1996a , 1996b ) and sandage ( 1996 ) , respectively .
the discovery of cepheids in ngc 4548 was reported by sakai ( 1996 ) with no distance values published yet . taken together with the earlier observation of ngc4571 the galaxy distances stretch over a wide range : ngc4571 with @xmath9 mpc ( pierce 1994 ) , m100 with @xmath10 mpc ( ferrarese 1996 ) , ngc4496 with @xmath11 mpc ( saha 1996a ) , ngc 4639 with @xmath12 mpc ( saha 1996b ) and ngc4639 with @xmath13 mpc ( sandage 1996 ) .
while the good resolvability of the galaxy ngc4571 ( sandage & bedke 1988 ) quite probably flags this object as being in the foreground of the major part of the virgo cluster , the other galaxies are believed to be true virgo spirals as argued by the authors of the cepheid observation papers ( see also tammann 1996 ) .
this highlights again the large extent of the galaxy infall region of the virgo cluster and the necessity of obtaining more precise information on the location of such galaxies with respect to the structure of the virgo cluster . in this paper
we therefore like to discuss how galaxy targets for cepheid measurements can be selected , for which the relative position to the center of virgo are known more precisely .
the selection of these targets is based on the fact that interaction effects of the interstellar medium with the intracluster gas of virgo can be observed for some cluster spirals .
in particular the two galaxies ngc4548 and ngc4501 are observed to be stripped off the interstellar gas ( cayatte 1990 , 1994 ) . in connection with the knowledge of the gas density distribution in the virgo cluster from rosat x - ray observations and theoretical considerations on the physics of the stripping process
, the location of these two galaxies can be confined to the surface of a certain isodensity shell of the intracluster medium around m87 .
the galaxy location can be well confined to a line - of - sight distance from m87 to within @xmath14 mpc .
this is about a factor of three more precise than in the case of m100 .
the virgo cluster as a whole is unfortunately quite complex , as observed in the optical galaxy distribution ( e.g. binggeli 1987 ) and in x - rays ( bhringer 1994 ) , and it is therefore difficult to get precise information on the line - of - sight location of m87 within the cluster . in the optical m87 is located near the center of the densest , northern part of the virgo cluster , designated cluster a in binggeli ( 1987 ) .
the x - ray observation of the virgo cluster in the rosat all sky survey provides some further clues on this problem , implying that m87 is probably centered in the halo of cluster a with some disturbance from the infall of the m86 galaxy group , as will be discussed in this paper .
this helps to relate the redshift of m87 to the recession velocity of the virgo cluster . in this way one can now relate the positions of the stripped galaxies at the edge of the intracluster gas halo around m87 to the recession velocity of m87 and the virgo cluster to derive the hubble parameter with largely reduced uncertainties as induced by the substructure in the virgo cluster .
we therefore propose that the two galaxies ngc4548 and ngc4501 should be chosen as targets for the search of cepheid variables to get a more reliable distance of the virgo cluster . in the following
we discuss the stripping of virgo cluster spiral galaxies and the density distribution of the gaseous x - ray halo of m87 in section 2 and derive the line - of - sight distances of ngc4548 and ngc4501 with respect to m87 . in section 3
we analyse the position of m87 with respect to the intracluster medium of the virgo cluster .
the results are discussed and summarized in section 4 . throughout the paper
we will assume a distance to the virgo cluster of 20 mpc and for the easy conversion of our results to other distance values we will give the scaling of the important quantities with distance by the parameter @xmath15 mpc .
as first proposed by gunn & gott ( 1972 ) spiral galaxies are stripped from their interstellar medium when they pass close enough to the core of a galaxy cluster and this effect was partly made responsible for the lack of late type galaxies in the densest regions of galaxy clusters ( e.g. dressler 1984 ) . in the virgo cluster
we can clearly observe this effect for example by a comparison of the neutral hydrogen content of spiral galaxies measured in the 21 cm radio line by van gorkum & kotanyi ( 1985 ) and cayatte ( 1990 ) and the distribution of the intracluster gas as observed in x - rays ( bhringer 1994 ) in the rosat all sky survey ( trmper 1993 , voges 1992 ) .
1 shows an overlay of the 21 cm measurement in 23 spirals by cayatte ( 1990 ) on top of the x - ray image of the virgo cluster from the rosat all sky survey .
the x - ray image displays the photon distribution detected in the energy range 0.4 to 2.4 kev and the image is smoothed with a variable gaussian filter .
as already noted by cayatte ( 1994 ) the galaxies closer to the center of the cluster marked by the giant elliptical galaxy m87
are almost devoid of neutral hydrogen whereas the extent of the neutral hydrogen gas disks increases with the distance from the cluster in the sample of galaxies observed . at an intermediate location
the galaxies show signs of distortion in their gas disks which can be interpreted as interaction effects of the gas disks with the hot , x - ray emitting intracluster medium ( icm ) of the virgo cluster .
this effect seems to be most evident for the two galaxies ngc 4501 and ngc 4548 .
2 shows a close - up picture of this region again as an overlay of the 21 cm image of the two spiral galaxies ngc 4501 and ngc 4548 to the x - ray surface brightness image of the icm of the virgo cluster from the rosat all sky survey .
the neutral hydrogen distribution in the two galaxies as seen in 21 cm is clearly asymmetrical .
the interstellar gas appears to be compressed at the side of the galaxies facing the virgo cluster center and the gas is combed backwards in the outer parts of the galaxies .
this is just what would be expected from the effect of ram pressure stripping .
the virgo cluster center marked by the galaxy m87 is by far the deepest gravitational potential in the region .
it is therefore certain that the galaxies are attracted by it and thus if they are on their first infall they are currently moving towards the cluster center .
therefore the side with which they are facing m87 ( located in the direction of the lower right corner in fig .
2 ) is the front - side with respect to their peculiar motion .
bernoulli forces and kelvin - helmholtz instabilities acting at the sides of the gas disks are responsible for the peanut shapes of the gas disks . for further details
see the images and data in cayatte ( 1990 , 1994 ) . for the x - ray image displayed in fig .
2 four obvious point sources in the region which are visible in fig . 1 have been removed to show only the angularly resolved structures which very probably correspond to features in the gas distribution of the virgo icm .
the icm appears actually quite patchy in this area .
it is striking that there are two surface brightness knots close to both of the being - stripped spiral galaxies . in both cases
the knots are offset from the galaxy in the direction of the motion .
they constitute surface density enhancements of about 30 - 40% above the background subtracted cluster emission .
they correspond to about 2@xmath16 enhancements in this heavily smoothed image ( gaussian used for smoothing is @xmath17 arcsec ) .
they have extents of up to 50 - 70 kpc with all the internal structure being smoothed - out . since the x - ray intensity is mostly a function of the emission measure in the line of sight ( and also because the intracluster gas is approximately in pressure equilibrium on these scales )
the intensity fluctuations have to correspond to density enhancements of the gas .
one can roughly estimate that local density enhancements with a scale of @xmath18 kpc and an amplitude of the order of a factor of two would produce the observed effect .
given the low significance and the badly resolved structure of these features we can only speculate .
but the features could indeed be explained by density enhancements in the icm due to the sweeping - up effect of the infalling galaxies . while the cold gas of the spirals is being stripped ,
the icm is piling up in front of the galaxies .
this would be a further confirmation of the fact that these two galaxies are strongly interacting with the intracluster gas of the virgo cluster . next we can discuss the ram pressure stripping effect quantitatively . as already discussed by gunn & gott ( 1972 , see also sarazin 1986 , cayatte 1994 ) this effect is determined by the balance of the ram pressure , @xmath19 , of the icm pushing the interstellar medium out off the spiral galaxy and the gravitational force with which the galaxy binds the interstellar medium .
the ram pressure depends on the density of the icm , @xmath20 , and the velocity of the galaxy , @xmath21 : @xmath22 the restoring force per unit area due to the gravity of the disk of the spiral galaxy is given by ( e.g. gott & gunn 1972 ) : @xmath23 where @xmath24 is the gravitational surface mass density and @xmath25 the surface density of the interstellar medium of the stripped spiral galaxy .
the 21 cm images of the spiral galaxies in the paper by cayatte ( 1990 ) are given out to the @xmath26 @xmath27 hydrogen column density contour .
therefore we calculate the stripping effect for this outer radius . for the mass density of the galactic disk we use the mass average inside this radius given by flat rotation curves and the assumption of keplerian orbits .
@xmath28 with this simplification we get @xmath29 and @xmath30 where @xmath31 is the column density of the neutral hydrogen ism as measured by the 21 cm emission and @xmath32 is the infall velocity of the spiral into the virgo cluster .
the information on the rotation curves can also be taken from 21 cm observations by guhathakurta ( 1988 ) .
for the infall velocity of the galaxies into the virgo cluster we take a value of 1200 km s@xmath2 which will be justified below .
the gas density in the center of the virgo cluster can be obtained from the observed x - ray surface brightness of the x - ray halo around m87 ( bhringer 1994 ; nulsen & bhringer 1995 ) with an approximate value as a function of radius of @xmath33 where @xmath34 is the electron density in the icm .
the properties are scaled by a virgo distance parameter @xmath35 mpc . taking the values for the rotational velocity from the rotation curves published by guhathakurta ( 1988 ) and the extents of the hydrogen disks to the @xmath26 @xmath27 contour from cayatte ( 1990 ) as listed in table 1 one
finds that the ram pressure is balanced by the gravitational force of the galaxy for an m87 distance of @xmath36 mpc for ngc 4501 and @xmath37 mpc for ngc 4548 .
the values we find are close to the values for the projected distances of the galaxies from m87 . for the calculations of the uncertainties
we have assumed that the considerations that went into the calculation of the pressure balance in the ram pressure effect may be uncertain by up to a factor of three . since we know
that ngc 4501 is falling in from the front and ngc 4548 from behind into the virgo cluster core , the distances to m87 are constraint to @xmath38 mpc for ngc 4501 and @xmath39 mpc for ngc 4548 .
likewise the line - of - sight distance difference between m87 and the stripped galaxies is @xmath40 mpc for ngc 4501 and @xmath41 mpc for ngc 4548 .
therefore within this galaxy stripping scenario the line - of - sight distance of the galaxies is fixed with respect to m87 with uncertainties of only one and two mpc , respectively .
this is a great improvement of the positional uncertainty in view of the use of these galaxies as probes for the cepheid distance scale .
[ cols= " < , < , < , < " , ]
with the analysis of the above observations we have in principle established the following important link : we know the line - of - sight positions of two late type galaxies quite precisely with respect to the location of m87 and we know that this galaxy is the center of the relaxed major mass component of virgo . now , to relate the recession velocity of m87 to the velocity of the main body of the virgo cluster we have to correct for the influence of the infalling galaxy group m86 .
with the positions of the galaxies ngc 4501 and ngc 4548 determined and a known redshift of the central part of virgo one has obtained a direct relation of distance and recession velocity which would allow to determine the hubble constant ( with the additional correction for the virgo infall ) in a more direct way than possible in the past .
it is of course crucial that both galaxies are of late type with a high enough star formation rate to make the search for cepheids feasible .
but since the stripping of the icm is just starting now and the galaxies still have a dense interstellar medium , there is no reason to expect a lower than normal population of cepheids in those galaxies .
both galaxies in particular ngc4501 have impressive dust lanes visible in the optical images implying that star formation should still be going on .
ngc 4501 and ngc 4548 are in fact the galaxies closest in projection to m87 for which one can hope to find cepheids for a distance measurement .
therefore we propose that these two galaxies should be made targets of cepheid searches with hst with high priority ( the observation of cepeids in ngc4548 was actually just reported by sakai 1996 while we were working on a revision of this paper ) .
in addition a better understanding of the dynamics within cluster a in virgo would be required .
of particular importance is a better determination of the masses of the halo regions of m87 and m86 .
an improved and more detailed analysis of the x - ray observations is in preparation by us .
a deeper analysis of the optical data is complicated by the fact that the galaxy density in the immediate vicinity of m87 is surprisingly low .
if one would naively assume that the galaxy density is proportional to the density of the intracluster gas , one would expect a much larger number of galaxies in the neighborhood of m87 .
galaxy cannibalism has been tentatively made responsible for this circumstance .
this lack of galaxies is so striking that for example the maximum of the projected galaxy density is much closer to m86 than to m87 ( see binggeli 1987 , schindler 1995 ) .
this uneven galaxy distribution makes it rather difficult to use the galaxy redshift distribution in the a cluster to get a more detailed picture of the m87-m86 merger .
we can further study the infall of the stripped galaxies in some more detail .
ngc 4501 has a recession velocity of 2283 indicating that it is falling in from the front , while ngc 4548 with a recession velocity of 483 is falling in from behind ( cayatte 1994 , heliocentric velocities ) .
m87 has for comparison a recession velocity of 1258 . from the analysis of the x - ray data of the rosat all sky survey in the virgo region ( bhringer et al .
1994 ) we get a rough estimate of the mass profile of the m87 halo region with values in the range @xmath42 .
the arrival velocity of a galaxy in such a potential at a radius of about 1 mpc is in the range of 1200 to 1900 .
here we have assumed that the potential of the cluster has been constant over the relevant time , that is at least for the last @xmath43gys , which may lead to a slight overestimate of the galaxy velocity .
the assumed value of the infall velocity in our ram pressure stripping analysis is therefore in the right range .
this analysis can also be improved when better data on the mass distribution in the m87 halo become available
. this will be possible with better information on the temperature distribution which will be obtained from planned asca observations .
there may also be other galaxies near the stripping radius that could be used as targets after more detailed studies , for which the present data do not show such clear evidence of stripping , like for example the galaxy ngc 4569 .
for the modeling of the stripping effect it would be useful to have even more detailed hi observations in those galaxies .
one should note that a good determination of the distance to virgo is not only important for a better knowledge of @xmath6 , but is also interesting in itself .
virgo has been the target of so many studies in particular in the field of environmental effects on the galaxy population that a proper distance scaling of all the studied physical processes would be a great advantage .
we thank gustav tammann , bruno binggeli and in particular renzo sancisi for fruitful discussions and helpful comments .
h.b . and s.s
are grateful for financial support by the german bmft through the verbundforschung for astronomy .
s.s . gratefully acknowledges the hospitality of the astronomical institute of the university of basel .
ferrarese , l. , freedman , w.l . , hill , r.j . ,
saha , a. , madore , b.f . ,
kennicut , r.c . ,
stetson , p.b . ,
ford , h.c . ,
graham , j.a . , hoessel , j.g . ,
han , m. , huchra , j. , hughes , s.m . ,
illingworth , g.d .
, kelson , d. , mould , j.r . , phelps , r. , silbermann , n.a . ,
sakai , s. , turner , a. , harding , p. , bresolin , f. , 1996 , , 464 , 568 freedman , w.l . , madore , b.f .
, mould , j.r . , hill , r. , ferrarese , l. , kennicutt , r.c .
, saha , a. , stetson , p.b . , graham , j.a . ,
ford , h. , hoessel , j.g . ,
huchra , j.p . ,
hughes , s.m .
, & illingwoth , g.d . , 1994 , , 371 , 757 mould , j. , huchra , j.p . ,
bresolin , f. , ferrarese , l. , ford , h.c . , freedman , w.l .
, graham , j. , harding , p. , hill , r. , hoessel , j.g . ,
hughes , s.m . , illingworth , g.d .
, kelson , d. , kennicutt , r.c . ,
madore , b.f . , phelps , r. , stetson , p.b . , & turner , a. , 1995 , , 449 , 413 tammann , g.a . ,
labhardt , l. , federspiel , m. , sandage , a. , saha , a. , macchetto , f.d . ,
panagia , n. , 1996 , in _ science with the hubble space telescope - ii _ , space telescope science institute publication , p. benvenuti , f.d .
macchetto ( eds . ) , 1996 | the measurement of precise galaxy distances by cepheid observations out to the distance of the virgo cluster is important for the determination of the hubble constant .
the virgo cluster is thereby often used as an important stepping stone .
the first hst measurement of the distance of a virgo galaxy ( m100 ) using cepheid variables provided a value for @xmath0 ( freedman 1994 ) .
this measurement was preceeded by a ground based study of the virgo spiral ngc4571 ( pierce 1994 ) formally providing @xmath1 km s@xmath2 .
these determinations rely on the accuracy with which the position of this observed spiral galaxy can be located with respect to the virgo cluster center .
this uncertainty introduces a major error in the determination of the hubble constant , together with the uncertainty in the adopted virgo infall velocity of the local group .
here we propose the use of spiral galaxies which show clear signs of being stripped off their interstellar medium by the intracluster gas of the virgo cluster as targets for the cepheid distance measurements .
we show that the stripping process and the knowledge of the intracluster gas distribution from rosat x - ray observations allow us to locate these galaxies with an at least three times higher precision with respect to m87 than in the case of other spirals like m100 .
the x - ray observations further imply that m87 is well centered within the densest part of the intracluster gas halo of the virgo cluster .
this seems to imply that m87 is approximately at rest in the central core of the cluster .
there remains the problem , however , that there is a velocity difference between m87 and the average velocity of the virgo galaxies in the larger halo region around m87 .
this may be explained by the disturbance introduced by the galaxy m86 , which is falling into the cluster with high velocity from behind .
the x - ray and optical observations imply that m86 is associated with a group of galaxies and a larger dark matter halo with a mass of up to about 10% of the mass of the m87 halo , which would account for the velocity offset of m87 with respect to the velocity average of about 100 to 150 km s@xmath2 .
the combination of these informations could enable us to locate the two stripped spiral galaxies quite precisely within the virgo cluster as a whole and could greatly improve the determination of the virgo cluster distance . |
in the standard electroweak model @xcite , the couplings of the leptonic charged and neutral currents to the gauge bosons are independent of the lepton generation . measurements of the lifetime , @xmath4 , and the leptonic branching fractions , @xmath5 , of the tau lepton provide a test of this universality hypothesis for the charged current .
the leptonic width of the tau lepton @xcite , @xmath6 where @xmath7e,@xmath8 , depends on the coupling constants of the tau lepton and the lighter lepton to the w boson , @xmath9 and @xmath10 , respectively .
here @xmath11 and @xmath12 are the masses of the tau lepton and the w boson .
the quantities @xmath13 , @xmath14 and @xmath15 are small corrections resulting from phase - space , radiation and the w propagator , respectively .
for the decay of the muon , @xmath16 , the same formula applies , with the muon mass and coupling , @xmath17 and @xmath18 , replacing those of the tau .
the comparison of the tau lifetime and leptonic branching fractions with the muon lifetime gives a direct measurement of the ratios @xmath19 and @xmath20 .
tau decays into hadrons are sensitive to the strong coupling constant , @xmath21 , at the tau mass scale . assuming universal coupling constants of the different lepton species to the w boson , the ratio of the hadronic width to the leptonic width , @xmath22 , can be expressed as : @xmath23 where @xmath24 contains the same corrections mentioned above .
@xmath22 is calculated in perturbative qcd @xcite : @xmath25 the quantities @xmath26 and @xmath27 are elements of the cabibbo - kobayashi - maskawa ( ckm ) quark mixing matrix @xcite ,
@xmath28 @xcite and @xmath29 @xcite describe electroweak radiative corrections and non - perturbative qcd contributions , respectively . the quantity @xmath30 is estimated as @xmath30 = 27.5 @xcite .
the perturbative theoretical uncertainty is taken to be @xmath30 = [email protected] .
this paper presents a measurement of the tau lepton lifetime with the l3 detector at lep , using data taken in 1995 at the @xmath31 pole .
furthermore , data from 1994 are re - analysed using an improved calibration and alignment of the central tracker .
the lifetime is measured from the decay length in three - prong tau decays and from the impact parameter of one - prong tau decays - prong tau lepton decay indicates a decay with @xmath32 charged particles in the final state . ] .
the results are combined with our previously published analyses on data collected from 1990 to 1993 @xcite .
previous measurements of the tau lepton lifetime have been reported by other experiments @xcite .
the l3 detector is described in ref .
this measurement is based primarily on the information obtained from the central tracking system , which is composed of a silicon microvertex detector ( smd ) @xcite , a time expansion chamber ( tec ) and a z - chamber .
the smd is made of two concentric layers of double - sided silicon detectors , placed at about 6 and 8 cm from the beam line .
each layer provides a two - dimensional position measurement , with a resolution of @xmath33 for normally incident tracks , in the directions perpendicular and parallel to the beam direction , denoted as @xmath34 and @xmath35 coordinates , respectively .
the tec consists of two coaxial cylindrical drift chambers with 12 inner and 24 outer sectors .
the sensitive region is between 10 and 45 cm in the radial direction , with 62 anode wires having a spatial resolution of approximately @xmath36 in the plane perpendicular to the beam axis .
the z - chamber , which is situated just outside the tec , provides a coordinate measurement along the beam axis direction .
for this measurement data collected in 1994 and 1995 are used , which correspond to an integrated luminosity of 49 pb@xmath37 and 31 pb@xmath37 , respectively . for efficiency and background estimates ,
monte carlo events are generated using the programs koralz @xcite for @xmath38 and , bhagene @xcite for @xmath39 , diag36 @xcite for @xmath40 , where @xmath41 is @xmath42 , @xmath43 , @xmath44 or @xmath45 , and jetset @xcite for @xmath46 .
the monte carlo events are passed through a full detector simulation based on the geant program @xcite , which takes into account the effects of energy loss , multiple scattering , showering and small time dependent detector inefficiencies .
these events are reconstructed with the same program used for the data .
the number of monte carlo events in each process is about ten times larger than the corresponding data sample .
tau lepton pairs originating from @xmath31 decays are characterised by two low multiplicity , highly collimated jets in the detector .
the selection of @xmath47 events is described in detail in ref .
@xcite ; here only a general outline is given . in order to have high - quality reconstruction of the tracks ,
events are accepted within a fiducial volume defined by @xmath48 , where the polar angle @xmath49 is given by the thrust axis of the event with respect to the electron beam direction .
the events must have at least two jets , and the number of tracks in each jet must be less than four .
the background from @xmath50 events is reduced by requiring the total energy deposited in the electromagnetic calorimeter to be less than @xmath51 of the centre - of - mass energy . to reduce background from @xmath52
the sum of the absolute momenta measured in the muon spectrometer must be less than @xmath53 of the centre - of - mass energy .
if muons are not reconstructed in the muon chambers they are identified by an energy deposit in the calorimeters which is characteristic of a minimum ionising particle .
if this is the case for one jet , the opposite jet is required to exhibit a hadronic signature .
this rejects dimuon as well as cosmic - ray events .
the cosmic - ray background is further reduced by requiring a scintillation counter hit within 5 ns of the beam crossing .
in addition , the distance of closest approach to the interaction point measured by the muon chambers must be less than two standard deviations of the resolution . following this procedure , @xmath54 and @xmath55 events are selected from the data collected in 1994 and 1995 , respectively . the selection efficiency in the fiducial volume is estimated to be 76% .
the purity of the tau pair sample is 98% .
for this measurement a high quality of the track reconstruction is essential .
a prerequisite is the control of the alignment between smd and tec and the drift time to drift distance calibration for the tec .
these calibrations , alignments and the estimation of resolution functions are performed with a clean sample of bhabha and dimuon events , where tracks are known to originate from a common vertex .
particular effort is invested in the individual alignment of each sensor of the smd and in the calibration of the boundaries of the tec sectors .
the procedure is described in detail in ref .
the performance of the track reconstruction is estimated from the distance between the two tracks at the vertex projected into the @xmath34 plane .
this quantity , called miss distance , is independent of the size and position of the @xmath56 interaction region .
the distributions of miss distance for bhabha and dimuon events collected during 1994 and 1995 are shown in figure [ fig : reso ] .
a gaussian function is fitted to both distributions , from which an intrinsic resolution @xmath57 and @xmath58 is estimated for 1994 and 1995 , respectively . to guarantee good tracks for the analysis ,
the following cuts are made : * number of hits in the tec @xmath59 .
this ensures a good curvature measurement . *
number of smd hits in the @xmath34 plane @xmath60 .
this criterion selects tracks for which the error in the extrapolation to the vertex is well described by @xmath61 .
* transverse momentum , @xmath62 .
tracks with lower momenta have a larger uncertainty in the extrapolation to the vertex due to multiple coulomb scattering in the smd .
* probability , @xmath63 , of the track fit larger than 1% .
this requirement rejects bad fits .
for three - prong tau decays , the decay vertex of the tau is reconstructed and its distance to the centre of the interaction region is measured .
the decay vertex is found from a minimisation with respect to the vertex coordinates @xmath64 of the following @xmath65 : @xmath66 in this equation , @xmath67 is the distance of closest approach of a track to the decay vertex coordinates .
the error , @xmath68 , is the quadratic sum of the intrinsic resolution , @xmath61 , and the error due to multiple coulomb scattering , @xmath69 .
figure [ fig : pc2 ] shows the distribution of the confidence level , @xmath63 , of the fit .
decay vertices with @xmath70 contain in most cases tracks with wrongly matched smd hits and are rejected .
the remaining distribution is flat , indicating a good description of the errors .
figure [ fig : dl ] shows the decay length distribution for the data collected in 1994 and 1995 .
only decay length values in the range @xmath71~\mbox{mm}$ ] with an error better than 5 mm are accepted for the lifetime determination .
the number of selected decays is @xmath72 and @xmath73 for 1994 and 1995 data , respectively . to obtain the average decay length ,
an unbinned maximum likelihood fit is applied to the observed decay length distribution .
the likelihood function is calculated from a convolution of an exponential @xmath74 , describing the tau decay time using the average decay length @xmath75 as a parameter , with a gaussian resolution function @xmath76 .
the likelihood function is written as : @xmath77 in this equation the product runs over the accepted three - prong tau decays @xmath78 .
the second term on the right hand side has been added to take into account background carrying no @xmath79 lifetime information .
the background fraction , @xmath80 , is estimated from monte carlo and fixed in the fit .
the likelihood function @xmath81 is evaluated from the convolution of a dirac delta function with the experimental resolution function .
the fit minimises @xmath82 .
average decay lengths of @xmath83 and @xmath84 are determined for data from 1994 and 1995 , respectively .
the results of the fit are represented by the solid lines in figure [ fig : dl ] .
the tau lifetime and average decay length are related through the following expression : @xmath85 using this equation and taking into account the effects of radiation , lifetimes of @xmath86 and @xmath87 are obtained for the two data sets , where the errors are statistical only . in order to check the decay length method , the analysis is repeated on a sample of monte carlo events .
the difference between the input tau lifetime and the result of the fit is assigned as a systematic error due to the method .
systematic effects due to the non - ideal description of the resolution function are estimated from a one @xmath88 variation of its parameters .
the deviation of the smd radius from its nominal value is measured using bhabha and dimuon events by minimising the track coordinate residuals from overlapping silicon sensors .
deviations of @xmath89 and @xmath90 are found in the data of 1994 and 1995 , respectively .
these results are compatible with zero , and their uncertainties are translated into a systematic error on the lifetime .
uncertainties in the fraction of background events carrying no lifetime information are estimated from a @xmath91 variation of the fraction .
finally , the systematic error due to the fit range is estimated from the combined data by a 10% variation of their lower and upper bounds .
this error also includes the effect of a variation of the cut on @xmath63 between 0.1 and 1.5% .
table [ tab : dlsys ] summarises the systematic errors for the decay length method .
the lifetime measurements from the @xmath92 three - prong decays from 1994 and 1995 are combined .
the systematic error due to the resolution function description is taken to be uncorrelated ; for the other errors a 100% correlation is assumed .
the result is @xmath93
a second measurement of the tau lifetime is obtained from the impact parameter in the plane perpendicular to the beam axis for one - prong tau decays .
the impact parameter is the distance of closest approach , @xmath94 , of the track to the tau lepton production point , which is estimated by the beam position . in order to be sensitive to the lifetime , a sign is given to the impact parameter , according to the position of the intersection between the track and the tau direction of flight with respect to the beam position .
the uncertainty on the impact parameter , @xmath95 , is described as the quadratic sum of the intrinsic detector resolution , @xmath61 , the size of the interaction region , @xmath96 , and a momentum dependent multiple coulomb scattering contribution , @xmath69 , @xmath97 where @xmath98 is the azimuthal angle in the plane perpendicular to the beam axis .
the average values for the interaction region size are determined from bhabha and dimuon data and listed in table [ tab : bspot ] .
in contrast to the decay length method a more complicated function for the description of the tau decay is expected here , since for a one - prong tau decay the decay vertex is _ a priori _ not known . from a monte carlo study of the impact parameter distribution at generator level
it is found that this function can be described in terms of three exponentials for positive and three exponentials for negative impact parameter values @xmath99 in this equation , @xmath100 represents the fraction of negative impact parameter values , which originate from an imperfect reconstruction of the tau flight direction .
the slopes of the exponentials , @xmath101 , contain the lifetime dependence of the distribution . as for the decay length method
, the lifetime is extracted from an unbinned maximum likelihood fit to the observed distribution .
the likelihood is now determined from the convolution of a double gaussian resolution function , which is obtained from bhabha and dimuon samples , with the function of eqn.(9 ) .
the fit also accounts for background carrying no lifetime information .
figure [ fig : ip ] shows the impact parameter distributions from tau decays collected in 1994 and 1995 .
impact parameters with a value in the range @xmath102~\mbox{mm}$ ] and an impact parameter error of less than @xmath103 are accepted for the measurement .
the fit yields a tau lifetime of @xmath104 and @xmath105 for the two data samples .
the errors are statistical only .
the method is checked on a monte carlo sample , from which the lifetime is determined in the same way as for the data .
the difference between the input tau lifetime and the result of the fit is assigned as a systematic error due to the method .
systematic effects due to the uncertainty of the resolution function are estimated from a variation of its parameters according to their errors , with correlations taken into account .
the beam spot size is varied according to its statistical errors .
the change in the central value is assigned as a systematic uncertainty .
the effect of the average smd radial position uncertainty is treated in the same way as in the decay length analysis .
the systematic effect due to the knowledge of the function @xmath106 is evaluated by taking into account its statistical uncertainty and its dependence on the tau lifetime in the range from 250 to 350 @xmath107 .
the uncertainty arising from the fraction of background events is estimated from a @xmath91 variation of this fraction . the systematic error induced by the choice of the fit ranges
is estimated from the combined data sample by a 10% variation of their bounds .
table [ tab : ipsys ] summarises the systematic errors for the impact parameter method .
the lifetime measurements from the @xmath108 one - prong decays from 1994 and 1995 are combined .
the systematic error due to the resolution function is taken to be uncorrelated .
for the other errors a 100% correlation has been assumed .
the result is @xmath109
the combination of the results obtained by the two methods with our previous ones @xcite yields @xmath110 correlations within the systematic errors are taken into account .
this result supersedes all previous results @xcite .
this value is in good agreement with the current world average @xcite . the measurements of the branching fractions , @xmath111 and @xmath112 @xcite together with this lifetime measurement and the muon lifetime @xcite yield @xmath113 and @xmath114 supporting the universality hypothesis . from the tau lifetime , the tau mass , the muon mass and muon lifetime , @xmath22
is found to be @xmath115 .
this corresponds to @xmath116 the first error is due to the errors of the tau lifetime and the ckm matrix elements @xcite .
the second error is the quadratic sum of the uncertainties resulting from the renormalisation scale , the term fourth order in @xmath21 , the electroweak corrections @xmath28 , and the non - perturbative correction , @xmath29 .
the renormalisation scale uncertainty is estimated following ref .
@xcite by a variation between 0.4 @xmath117 2.0 and is the dominant contribution to the error .
other contributions to the theory error as discussed in ref .
@xcite are not considered .
this result is in good agreement with other measurements of @xmath21 at the tau mass @xcite .
the value of @xmath118 is extrapolated to the @xmath31 mass using the renormalisation group equation @xcite with the four loop calculation of the qcd @xmath119-functions @xcite .
the result , @xmath120 = 0.1185 @xmath3 0.0019 @xmath3 0.0017 , is in good agreement with the current world average value @xcite .
we thank g. altarelli and a. kataev for discussions about the estimation of the theoretical uncertainty of @xmath22 . we wish to express our gratitude to the cern accelerator divisions for the excellent performance of the lep machine .
we acknowledge the contributions of the engineers and technicians who have participated in the construction and maintenance of this experiment .
s. glashow , nucl .
* 22 * ( 1961 ) 579 ; s. weinberg , phys .
* 19 * ( 1967 ) 1264 ; a. salam , elementary particle theory , edited by n. svartholm ( almqvist and wiksell , stockholm , 1968 ) , p. 367 .
. l3 collab .
, o. adriani , physics reports * 236 * ( 1993 ) 1 ; m. biasini , proceedings of the third workshop on tau lepton physics , montreux , switzerland 1994 , l. rolandi edt .
b ( proc . suppl . )
* 40 * ( 1995 ) 331 .
aleph collab .
, r. barate , ( 1997 ) 362 ; + delphi collab . , p. abreu , ( 1996 ) 448 ; + opal collab . , g. alexander , ( 1996 ) 341 ; + cleo collab . , r. balest , ( 1996 ) 402 ; + sld collab . , k. abe , ( 1995 ) 4828 ; + jade collab.,c .
kleinwort , ( 1989 ) 7 ; + mark2 collab . ,
d. amidei , ( 1988 ) 1750 ; + tasso collab . , w. braunschweig , ( 1988 ) 331 ; + hrs collab .
, s. abachi , ( 1987 ) 2519 ; + argus collab . , h. albrecht , ( 1987 ) 580 ; + mac collab . , h. band , ( 1987 ) 415 .
the l3 detector simulation is based on geant version 3.15 .
+ see r. brun , `` geant 3 '' , cern dd / ee/84 - 1 ( revised ) , september 1987 .
+ the gheisha program ( h. fesefeldt , rwth aachen report pitha 85/02 ( 1985 ) ) is used to simulate hadronic interactions .
m.acciarrip.achard o.adriani m.aguilar-benitez j.alcaraz g.alemannij.allabya.aloisio m.g.alviggig.ambrosih.anderhub v.p.andreevt.angelescuf.anselmoa.arefiev t.azemoon t.aziz p.bagnaiaa.bajo l.baksaya.balandras s.banerjee sw.banerjee a.barczyk r.barillre l.barone p.bartalini m.basiler.battistona.bay f.becattiniu.beckerf.behnerl.bellucci r.berbeco j.berdugo p.berges b.bertuccib.l.betevs.bhattacharyam.biasinia.biland j.j.blaising s.c.blyth g.j.bobbink a.bhml.boldizsarb.borgia d.bourilkovm.bourquins.braccinij.g.bransonv.brigljevic f.brochu a.buffinia.buijsj.d.burgerw.j.burgerx.d.cai m.campanellim.capellg.cara romeog.carlinoa.m.cartacci j.casausg.castellinif.cavallarin.cavallo c.cecchi m.cerradaf.cesaroni m.chamizoy.h.chang u.k.chaturvedi m.chemarina.chen g.chen g.m.chen h.f.chen h.s.cheng.chiefari l.cifarellif.cindoloc.civinini i.clarer.clare g.coignet a.p.colijnn.colino s.costantini f.cotorobaib.cozzoni b.de la cruza.csillings.cucciarelli t.s.dai j.a.van dalen r.dalessandro r.de asmundisp.dglon a.degr k.deiters d.della volpe p.denes f.denotaristefania.de salvo m.diemoz d.van dierendonckf.di lodovicoc.dionisi m.dittmara.domingueza.doriam.t.dovad.duchesneau d.dufournaud p.duinker i.duranh.el mamounia.engler f.j.eppling f.c.ern p.extermann m.fabre r.faccinim.a.falagans.falcianoa.favaraj.fay o.fedinm.felcinit.ferguson f.ferronih.fesefeldt e.fiandrinij.h.field f.filthautp.h.fisheri.fiskg.forconi
l.fredjk.freudenreichc.furettayu.galaktionovs.n.ganguli p.garcia-abiam.gataullins.s.gaus.gentilen.gheordanescus.giaguz.f.gongg.grenier o.grimm m.w.gruenewald m.guida r.van gulikv.k.gupta a.gurtul.j.gutayd.haasa.hasan d.hatzifotiadout.hebbekera.herv p.hidasj.hirschfelderh.hofer g. holzner h.hooranis.r.houy.hu i.iashvilib.n.jin l.w.jonesp.de jongi.josa-mutuberrar.a.khan m.kaurm.n.kienzle-focaccid.kimj.k.kimj.kirkbyd.kissw.kittela.klimentov a.c.kniga.koppv.koutsenko m.krber r.w.kraemerw.krenz a.krger a.kunin p.ladron de guevarai.laktinehg.landik.lassila-perinim.lebeaua.lebedevp.lebrunp.lecomte p.lecoq p.le coultre h.j.leej.m.le goffr.leiste e.leonardip.levtchenkoc.li s.likhoded c.h.linw.t.linf.l.lindel.listaz.a.liuw.lohmanne.longo y.s.lu k.lbelsmeyerc.luci d.luckeyl.lugnier l.luminariw.lustermannw.g.ma m.maityl.malgeria.malinin c.maad.mangeolj.mans p.marchesini g.marian j.p.martin f.marzano g.g.g.massaro k.mazumdarr.r.mcneil s.melel.merola m.meschini w.j.metzgerm.von der meya.mihulh.milcentg.mirabelli j.mnichg.b.mohanty p.molnarb.monteleoni r.moore t.moulikg.s.muanzaf.muheima.j.m.muijsm.musy m.napolitanof.nessi-tedaldih.newman t.niessena.nisatih.nowak g.organtinia.oulianov c.palomaresd.pandoulas s.paolettip.paoluccir.paramatti h.k.parki.h.parkg.pascaleg.passalevas.patricelli t.paulm.pauluzzic.pausf.paussm.pedaces.pensottid.perret-gallix b.petersend.piccolo f.pierella m.pierip.a.pirou e.pistolesiv.plyaskin m.pohl v.pojidaevh.postemaj.pothiern.produitd.o.prokofiev d.prokofiev j.quartierig.rahal-callotm.a.rahaman p.raics n.rajar.ramelli p.g.rancoitaa.raspereza g.ravenp.razisd.ren m.rescignos.reucroftt.van rhees.riemannk.rilesa.robohmj.rodinb.p.roel.romero a.rosca s.rosier-lees j.a.rubio d.ruschmeierh.rykaczewski s.saremi s.sarkarj.salicio e.sanchezm.p.sandersm.e.sarakinosc.schferv.schegelskys.schmidt-kaerstd.schmitz h.schopperd.j.schotanusg.schwering c.sciaccad.sciarrino a.seganti l.servolis.shevchenkon.shivarovv.shoutko e.shumilov a.shvorobt.siedenburgd.sonb.smithp.spillantini m.steuerd.p.stickland a.stone b.stoyanova.straessnerk.sudhakarg.sultanovl.z.sunh.suter j.d.swainz.szillasit.sztaricskai x.w.tangl.tauscherl.taylorb.tellili c.timmermanssamuel c.c.ting s.m.ting s.c.tonwar j.tth c.tullyk.l.tungy.uchidaj.ulbricht e.valente g.vesztergombii.vetlitsky d.vicinanza g.viertel s.villam.vivargent s.vlachosi.vodopianov h.vogelh.vogt i.vorobiev a.a.vorobyov a.vorvolakosm.wadhwaw.wallraff m.wangx.l.wang z.m.wanga.weberm.weberp.wienemannh.wilkenss.x.wus.wynhoff l.xia z.z.xu j.yamamoto b.z.yang c.g.yang h.j.yangm.yangj.b.yes.c.yeh an.zaliteyu.zalitez.p.zhang g.y.zhur.y.zhua.zichichif.ziegler g.zilizim.zller .
miss distance distributions from 1994 ( left ) and 1995 ( right ) ; bhabha and dimuon events are shown in linear scale ( upper ) and logarithmic scale ( lower ) .
dots are data and the solid line is the result of a fit with a gaussian . , title="fig:",scaledwidth=49.0% ] miss distance distributions from 1994 ( left ) and 1995 ( right ) ; bhabha and dimuon events are shown in linear scale ( upper ) and logarithmic scale ( lower ) .
dots are data and the solid line is the result of a fit with a gaussian . , title="fig:",scaledwidth=49.0% ] miss distance distributions from 1994 ( left ) and 1995 ( right ) ; bhabha and dimuon events are shown in linear scale ( upper ) and logarithmic scale ( lower ) .
dots are data and the solid line is the result of a fit with a gaussian . , title="fig:",scaledwidth=49.0% ] miss distance distributions from 1994 ( left ) and 1995 ( right ) ; bhabha and dimuon events are shown in linear scale ( upper ) and logarithmic scale ( lower ) .
dots are data and the solid line is the result of a fit with a gaussian . , title="fig:",scaledwidth=49.0% ] | the tau lepton lifetime is measured with the l3 detector at lep using the complete data taken at centre - of - mass energies around the z pole resulting in @xmath0 .
the comparison of this result with the muon lifetime supports lepton universality of the weak charged current at the level of six per mille .
assuming lepton universality , the value of the strong coupling constant , @xmath1 is found to be @xmath2 = 0.319 @xmath3 0.015 @xmath3 0.014 . |
solar - like oscillations of giant stars were firstly confirmed in a few stars @xcite .
and then the presence of radial and non - radial solar - like oscillations in a large sample of red giants was presented by @xcite .
although only a very limited number of modes ( @xmath8 3 ) are likely to be observed in solar - like oscillations due to geometrical cancellation effects , the low - degree p - modes of the oscillations can penetrate deep into the interior of stars @xcite , and each low - degree p - mode carries unique information about the stellar interior .
thus , asteroseismology has the capability to probe the interior of stars and to determine stellar fundamental parameters @xcite .
however , for most of red giants , only the mean large - frequency separation ( ) and the frequency of maximum seismic amplitude ( ) have been obtained so far .
the oscillation frequencies and of a large sample of red giants were firstly extracted by @xcite from the data observed by the first _ convection rotation and planetary transits ( corot ) _
@xcite 150-day long run in the direction of the galactic centre ( lrc01 ) .
they found that there is a dominant peak in the distributions of and , respectively , and that there is a tight relation between and , i.e. @xmath9 where @xmath10 and @xmath6 are constant parameters . using the data observed by _
corot _ in the opposite direction of lrc01 ( lra01 ) , @xcite found similar results , except for the locations of the dominant peaks and the values of the parameters @xmath10 and @xmath6 . by making use of the data observed by
mission @xcite , @xcite and @xcite also obtained very similar results .
moreover , the values of and of red giants of clusters ngc 6811 , ngc 6819 , and ngc 6791 have also been extracted by @xcite and @xcite . for an ensemble of stars ,
the and can be applied to investigate many interesting questions about the galaxy or stellar clusters @xcite .
these studies advanced our understanding of the theory of stellar structure and evolution .
furthermore , it has been identified that the non - uniform distributions of and of red giants result from red - clump [ core - helium - burning(cheb ) ] stars @xcite .
the dominant peak in the distributions of and , which is located around 30 and 4 for and , respectively , is mainly composed of the stars that are close to the zero - age horizontal branch ( zahb ) @xcite . stars with @xmath11 40 are mainly red - clump stars with mass larger than 2.0 @xcite . these stars could be attributed to the secondary red clump ( src ) predicted by @xcite @xcite .
moreover , @xcite showed that the src stars could become a hump in the distribution roughly between 70 and 100 [ see fig .
[ fig1 ] or the figure 2 in @xcite ] .
@xcite also showed that the distributions of and observed by _
corot _ have a complex structure .
the src has been identified in the colour - magnitude diagrams of some intermediate - age star clusters , such as ngc 419 in the small magellanic cloud @xcite and ngc 1751 in the large magellanic cloud @xcite .
although the origin of the src in clusters is still an open question @xcite , the src should be present in all galactic fields containing stars with an age of about 1 gyr @xcite .
the src is believed to be made of stars whose mass is slightly more massive than the critical mass ( ) that is the maximum mass allowing helium flash to occur @xcite .
stellar models with @xmath12 can be generally evolved from zero - age main sequence ( zams ) to asymptotic giant branch ( agb ) without interruption . however , the models with @xmath5 are hardly computed through the violent helium flash .
thus , the stars with @xmath12 should be more suitable to study solar - like oscillations of cheb stars and the effects of convective - core overshooting and other physical processes on the oscillations in details than the stars with @xmath5 .
in addition , the astroseismic detection and study on src stars may provide additional constraints on the star - formation history in the galaxy and allow us to determine the value of . in this paper
, we mainly focused on the global characteristics of solar - like oscillations of src stars and on the effects of convective - core overshooting , mixing - length parameter , and metallicity on the characteristics .
the paper is organized as follows .
we show our stellar models and population - synthesis method in section 2 .
we present the results in section 3 and discuss and summarize them in section 4 .
we used the stellar evolution code of @xcite to calculate stellar evolution models .
the code has been updated with the recent input physics over the last three decades @xcite .
the equation of state of @xcite as modified by @xcite and standard mixing - length theory are used in the code . * the value of 2.0 for the mixing - length parameter ( @xmath2 )
was calibrated against the sun*. the convective - core overshooting is treated as the prescription of @xcite . the value of 0.12 for the overshooting parameter @xmath4 is chosen to match the properties of @xmath13 aurigae binaries which span a mass range of 2.5 to 7.0 @xcite .
the initial hydrogen and helium abundances are assumed to be functions of the metallicity as follows : @xmath14 and @xmath15 .
the efficiency of @xcite mass - loss rate is set to 0.4 for all models .
the opacity table for the metallicity is compiled by @xcite from @xcite and @xcite .
however , element diffusion for helium and metals is not taken into account .
all models with @xmath12 were evolved from zams to agb stage without interruption . for the low - mass stars ( @xmath5 )
we were unable to compute through the he - flash .
instead , we constructed zahb models using the prescription of @xcite .
these zahb models were evolved up to the agb stage .
it should be noticed that some errors could be introduced in the post he - flash evolution of these constructed zahb models @xcite .
in addition , for a star given mass , radius and effective temperature , the theoretical and were calculated by using scaling equations @xcite @xmath16 and @xmath17 here , the value of 134.6 for @xmath18 is obtained from the gong data @xcite .
the accuracy of these estimates is good to within 5% @xcite . in order to obtain the distributions of and of src stars , we calculated single - star stellar population ( ssp ) .
stellar samples are generated by the monte carlo simulation .
the basic assumptions for the simulations are as follows .
( i ) star formation rate ( sfr ) is assumed to be a constant .
( ii ) the age - metallicity relation is taken from @xcite or a constant metallicity is assumed for simplicity .
( iii ) the lognormal initial mass function ( imf ) of @xcite is adopted .
in order to obtain the and of the populations of src stars , we firstly generated a zams sample with mass between 1.80 and 5.0 according to the imf of @xcite .
then we computed the evolutions of the models in the sample from high to low mass until calculations were interrupted by a violent he - flash at the tip of first giant branch ( fgb ) . for a given @xmath3
, we define as the mass for which the at the beginning of quiescent helium burning [ ` horizontal branch ( hb ) ' .
hereafter , the ` hb ' also refers to the similar evolutionary stage of models with @xmath7 in the hertzsprung - russell diagram . ] reaches a maximum and he - core mass ( ) reaches a minimum .
the luminosity of the model with @xmath19 at the fgb tip almost reaches a minimum .
the value of the he - core mass is about 0.33 for @xmath20 0.02 and about 0.34 for @xmath20 0.004 .
the values of in table [ tab1 ] for different metallicities are estimated from the tracks calculated at an interval less than 0.005 .
these values are almost not affected by the mixing - length parameter @xmath2 but can be significantly affected by overshooting parameter @xmath4 . the stars with mass slightly less than ( for @xmath20 0.02 , the mass is between about 1.70 and 2.01 ) ignite helium under semi - degenerate conditions at core masses between @xmath21 0.33 and 0.45 ( _ hereafter referred to as semi - degenerate stars _ ) . however , for the stars with lower mass , helium is ignited in the degenerate core with mass about 0.45 ( _ hereafter referred to as degenerate stars _ ) @xcite .
.the values of the critical mass ( in ) estimated from our calculations with @xmath2 = 2.0 and @xmath4 = 0.12 . [ cols="<,<,<,<,<,<,<",options="header " , ] [ tab2 ] figure [ fig14 ] shows the parameters @xmath10 and @xmath6 as a function of population age . for the populations with @xmath3 = 0.02
, the value of @xmath10 increases from about 0.2 to 0.3 when the age of populations increases from about 0.2 to 9.0 gyr .
however , the value of parameter @xmath6 is about 0.78 and almost unchanged when age @xmath22 1.2 gyr and is around 0.764 but slightly decreases with age when age @xmath11 2.0 gyr . when 1.2 gyr @xmath22 age @xmath22 2.0 gyr , the value of @xmath6 decreases with age .
recent observations show that the intermediate - age star clusters may be composed of stars with a non - uniform age [ a spread in age of about 200 - 300 myr @xcite ] .
our calculations show that a spread in age of 200 myr can significantly affect the parameters @xmath10 and @xmath6 of the populations with age less than about 2 gyr but hardly affects those of the populations with age @xmath11 2 gyr .
the spread leads to an increase in @xmath6 but a decrease in @xmath10 for the populations with age less than about 1.2 gyr and a decrease in @xmath6 for the populations with age between about 1.2 and 2.0 gyr .
for the populations with the spread in age , the value of @xmath6 decreases fast from about 0.81 to 0.77 when the age of the populations increases from 0.2 to 2.0 gyr .
moreover , our calculations show that the characteristics of @xmath10 and @xmath6 of populations with @xmath3 = 0.03 , 0.04 and 0.05 are almost the same as those of populations with @xmath3 = 0.02 . however , the value of @xmath10 of populations with @xmath3 = 0.004 is larger than that of populations with @xmath3 = 0.02 .
but the value of @xmath6 of populations with @xmath3 = 0.004 is smaller than that of populations with @xmath3 = 0.02 . when the age of populations is larger than about 2 gyr , the value of @xmath6 for the populations with 0.02 @xmath23 0.04 is located between about 0.77 and 0.76 ; however , that for populations with @xmath3 = 0.004 is located between about 0.76 and 0.75 .
the value of @xmath6 of populations with @xmath20 0.01 is located between that of populations with @xmath20 0.02 and populations with @xmath20 0.004 .
furthermore , our calculations show that a change in the mixing - length parameter does not significantly affect the parameters @xmath10 and @xmath6 .
however , the decrease in @xmath4 can lead to a small increase in @xmath6 and decrease in @xmath10 for the populations with @xmath12 .
the value of parameters @xmath10 and @xmath6 is about 0.20 and 0.79 for ngc 6811 ( @xmath211.0 gyr ) , 0.24 and 0.77 for ngc 6819 ( @xmath212.5 gyr ) , and 0.28 and 0.75 for ngc 6791 ( @xmath218.5 gyr ) , respectively @xcite .
the characteristics of the @xmath10 and @xmath6 of populations with the spread in age of 200 myr are more consistent with these obtained by @xcite than those of populations with a uniform age .
we computed the stellar populations with @xmath2 = 2.7 , @xmath4 = 0.12 , and @xmath3 = 0.004 , 0.01 , 0.02 , 0.03 , 0.04 and 0.05 .
however , the metallicity of the populations in the galaxy is believed to change with age .
we estimated the metallicity from the age - metallicity relation of @xcite . by interpolating
, we obtained the cheb populations with age between 0 and 9 gyr and metallicity changing with age .
the distributions of and of the populations are shown in fig .
[ fig15 ] .
the dominant peak is located about 30 and 3.9 in the and distribution , respectively .
the src peak is situated at around 78 and 7 in the and distribution , respectively .
the shoulder is located between about 43 and 54 in the distribution .
however , there is not a significant shoulder in the distribution .
moreover , there is a deficit in the distribution between about 9 and 20 and in the distribution between around 1.2 and 2.5 , which is caused by the convective acceleration of cheb stars .
the peak in the histogram about 7 and in the histogram around 1 is mainly composed of the cheb stars with @xmath12 4.0 and the early agb stars .
this peak exists in the distributions of and observed by the _
the deficit in the histogram between about 10 and 20 and in the histogram between around 1.5 and 2.5 observed by _ corot _
@xcite might be partly caused by the convective acceleration .
the value of obtained from our definition is slightly different from that obtained from the definition of @xcite who defines as the mass for which the luminosity at the fgb tip reaches a minimum and the core mass at he ignition is about 0.33 .
according to @xcite definition , we obtained the value of is 1.994 for @xmath20 0.02 , @xmath24 2.0 , and @xmath25 0.12 .
the value is consistent with that obtained by @xcite .
our calculations show that the maximum of of the model with @xmath19 1.994 is about 76 on the ` hb ' , which is less than the value of 78 of the model with @xmath19 2.01 .
the src peak can appear in the distributions of and of populations with @xmath0 0.02 in our simulations .
the location of the src peak can be affected by the parameters @xmath26 , and @xmath4 .
an increase in @xmath2 or decrease in @xmath3 or @xmath4 can lead to an increase in the frequency of the location of the src peak .
the change in @xmath3 or @xmath2 can significantly affect the effective temperature of the populations of src stars rather than the age of the populations and the value of for @xmath0 0.02 .
however , the variation in @xmath4 can significantly affect the value of and the age of the populations rather than the effective temperature . according to the formulas ( [ eq2 ] ) and ( [ eq3 ] )
, the mass of a star can be estimated from its , and effective temperature @xmath27 . if the frequency of the location of the src peak can be obtained from observations , which may aid in determining the value of and calibrating the overshooting parameter @xmath4 .
moreover , the src peak is mainly composed of cheb stars with mass between about and + 0.2 whose he - core masses are between about 0.33 and 0.36 . because the mass range of the src peak stars is very limited , the src peak in the histograms of the and may aid in determining the star - formation rate in the galaxy or star clusters .
the src peak is considerable in our simulations . comparing with the dominant peak , it is , however , insignificant .
if the observed src stars have different @xmath26 and @xmath4 , the src peak might become inconsiderable .
the appearance of the shoulder in the histogram of is related to the convective acceleration of stars with mass @xmath28 and the of semi - degenerate stars decreasing with mass . the degenerate and semi - degenerate zero - age models are always constructed from a non - degenerate model by mass loss .
the and of these models are mainly dependent on the he - core mass of the zero - age models rather than other parameters .
we do not know the exact value of the he - core mass of stars with @xmath5 when they reach the ` hb ' . a small change in the he - core mass can lead to an obvious variation in the of semi - degenerate stars .
thus the distributions of and of semi - degenerate stars may be affected by the construction method of the zahb models .
if the of semi - degenerate stars decreases with mass too fast , the increase in the number of semi - degenerate stars with decreasing can not counteract the decrease in the number of stars with @xmath7 between about 40 and 50 , the convective acceleration peak will appear in the simulated histogram of of populations including semi - degenerate and degenerate stars , while the upper boundary of the shoulder location will move to a lower frequency , i.e. the shoulder might be replaced by a narrower shoulder plus a convective acceleration peak .
the frequency of the upper boundary of the shoulder or the location of convective acceleration peak is mainly determined by the frequency where the convective acceleration of src stars begins .
the value of this frequency can be affected by the parameters @xmath26 and @xmath4 .
for example , for populations with @xmath20 0.04 , the value is around 54 for @xmath24 2.7 but is about 47 for @xmath24 2.0 .
however , the frequency of the lower boundary of the shoulder is mainly determined by the degenerate stars .
the effect of @xmath26 and @xmath4 on the lower boundary is harder to study by evolving stellar models than that on the upper boundary .
a detailed statistical study on the red ginat stars with observed @xmath11 40 would provide additional constraints on the theory of structure and evolution of semi - degenerate and src stars . the src and semi - degenerate stars with the same have different , i.e. the values of parameters @xmath10 and @xmath6 for src stars are different from those for semi - degenerate stars , which leads to a small difference between the and distributions .
for example , the shoulder is less significant in the distribution than that in the distribution .
the parameter @xmath10 of cheb populations increases with the age of populations in the range of about 0.20 @xmath29 0.33 .
the value of the parameter @xmath6 for populations with @xmath7 ( age @xmath30 1.1 gyr ) is obviously larger than that for the populations with @xmath5 .
the value of @xmath6 for old populations is approximately equal . in general
, a spread in the age of populations leads to a decrease in @xmath10 and increase in @xmath6 for the populations with @xmath12 but results in a small increase in @xmath10 and decrease in @xmath6 for the populations with @xmath5 . when the star with @xmath19 and @xmath20 0.02 evolved into the ` hb ' , its he - core mass is about 0.33 . when the he - core mass increases from about 0.33 to 0.36 , the energy production of he burning increases slightly but that of h - shell burning decreases and reaches a minimum , the total energy production is almost constant .
the and of this model is almost unchanged in this phase .
when the stars with mass slightly larger than ( roughly between and + 0.2 ) and @xmath20 0.02 arrive at the ` hb ' , their he - core masses are between 0.33 and 0.36 and their and are almost equal .
our simulations show that these stars can form an src peak in the histograms of and of the cheb populations .
the src peak can also appear in the histograms of and of the cheb populations with @xmath20 0.03 , 0.04 , and 0.05 .
however , it does not exist in the histograms of and of the cheb populations with @xmath20 0.01 and 0.004 .
the src peak is present in the histograms of and of the cheb populations calculated according to the age - metallicity relation of @xcite too .
the src peak location can be affected by the mixing - lenght parameter @xmath2 , core overshooting parameter @xmath4 and metallicity @xmath3 .
an increase in @xmath2 or decrease in @xmath4 or @xmath3 leads to a movement of the peak location towards a higher frequency .
the change in @xmath4 can significantly affect the value of but hardly affects the effective temperature of the cheb populations , which is different from the effects of @xmath2 and @xmath3 .
the src peak would aid in determining the value of and calibrating the overshooting parameter @xmath4 .
moreover , the convective acceleration in the stars with @xmath12 and the of semi - degenerate stars decreasing with mass result in the appearance of a shoulder in the distribution .
the convective acceleration of stars with @xmath5 leads to the deficit in the histogram between about 9 and 20 and in the histogram between about 1.5 and 2.5 .
this work was supported by china postdoctoral science foundation funded project 20100480222 , the ministry of science and technology of the people s republic of china through grant 2007cb815406 , the nsfc though grants 10773003 , 10933002 , 11003003 , 10963001 , and the project of science and technology from the ministry of education ( 211102 ) .
de ridder j. , barban c. , carrier f. , mazumdar a. , eggenberger p. , aerts c. , deruyter s. , vanautgaerden j. , 2006 , a&a , 448 , 689 de ridder j. et al .
, 2009 , nature , 459 , 398 dziembowski w. a. , goode , p. , 1997 ,
a&a , 317 , 919 eggleton p. p. , 1971 ,
mnras , 151 , 351 eggleton p. p. , 1972 ,
mnras , 156 , 361 eggleton p. p. , 1973 ,
mnras , 163 , 279 eggleton p. p. , faulkner j. , flannery b. p. , 1973
, a&a , 23 , 325 frandsen s. et al . , 2002 ,
a&a , 394 , l5 frandsen s. et al . , 2007 , a&a , 475 , 991 pols o. r. , tout c. a. , eggleton p. p. , han z. , 1995 , mnras , 274 , 964 pols o. r. , schroder k. p. , hurley j. r. tout c. a. , eggleton p. p. , 1998 ,
mnras , 298 , 525 reimers d. 1975 , memoires of the societe royale des sciences de liege , 8 , 369 rocha - pinto h. j. , maciel w. j. , scalo j. , flynn c. , 2000 , a&a , 358 , 850 yang w. , bi s. , 2007 , a&a , 472 , 571 yang w. , li z. , meng x. , bi , s. , 2011a , mnras , 414 , 1769 yang w. , meng x. , 2009 , pasj , 61 , 1399 yang w. , meng x. , 2010 , new astron . , 15 , 367 yang w. , meng x. , bi , s. , tian , z. , li , t. , liu , k. , 2011b , apj , 731 , l37 yang w. , meng x. , li z. , 2010 , mnras , 409 , 873 | we calculated the populations of core - helium - burning ( cheb ) stars and found that the secondary red clump ( src ) stars can form an src peak in the distributions of the frequency of maximum seismic amplitude ( ) and mean large - frequency separation ( ) of cheb stars when metallicity @xmath0 0.02 .
the and of cheb stars are dependent not only on he core mass but on h - shell burning .
the src peak is composed of the cheb stars with mass roughly between the critical mass and @xmath1 while he core mass is between about 0.33 and 0.36 .
the location of the src peak can be affected by the mixing - length parameter @xmath2 , metallicity @xmath3 , and overshooting parameter @xmath4 . a decrease in @xmath2 or
increase in @xmath3 or @xmath4 leads to a movement of the src peak towards a lower frequency .
however , the change in @xmath3 and @xmath2 only slightly affects the value of but the variation in @xmath4 can significantly affects the value of .
thus the src peak might aid in determining the value of and calibrating @xmath4 .
in addition , the effects of convective acceleration of src stars and the of ` semi - degenerate ' stars decreasing with mass result in the appearance of a shoulder between about 40 and 50 in the distribution . however , the convective acceleration of stars with @xmath5 leads to the deficit in the distribution between about 9 and 20 .
moreover , the value of the parameter @xmath6 of the relation between and for the populations with @xmath7 is obviously larger than that for the populations with @xmath5 .
[ firstpage ] stars : evolution ; stars : late - type ; stars : oscillations . |
by improving a lattice action , one hopes that calculations at a fixed lattice spacing will be closer to their continuum values than with the original action , and similarly that calculations of a fixed accuracy can be performed on coarser lattices than those required , for example , by the wilson action . for the present work
, we consider an action which has classical errors beginning only at @xmath2 , and which is also tadpole - improved@xcite .
this action has been used to compute the spectrum of light hadron masses@xcite , and it was found that hadron mass ratios are considerably improved compared to wilson action results at comparable lattice spacings . in this work
we address the question of whether matrix elements are similarly improved by considering the vector current renormalization factor and vector meson decay constant .
we use an improved gauge field action which involves a sum over 1@xmath32 plaquettes as well as 1@xmath31 plaquettes@xcite , and an improved fermion action which contains next - nearest - neighbour interactions@xcite , @xmath4 \nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\!\!\ !
-\frac{\kappa}{6u_0}\sum_{x,\mu } \left[\bar{\psi}(x)(2-\gamma_{\mu } ) u_{\mu}(x)u_{\mu}(x+\mu)\psi(x+2\mu)\right .
\!\!\!\!\!\!\!\!\!\!\nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\!\!\
! \left .
+ \bar{\psi}(x+2\mu)(2+\gamma_{\mu } ) u_{x}^{\dagger}(x+\mu)u_{\mu}^{\dagger}(x)\psi(x)\right]~.\end{aligned}\ ] ] the tadpole factor@xcite is @xmath5 .
the presence of three timesteps in a single term of the action introduces artifacts@xcite , such as oscillations in meson mass functions near the source .
however on our lattices , the plateau itself is not altered .
in the continuum , the @xmath0 meson decay constant @xmath6 is defined as follows : @xmath7 on a lattice , matrix elements of the vector current get renormalized by a factor @xmath8 , @xmath9 where the dots represent finite lattice spacing effects . for the improved action ,
the @xmath10 terms vanish identically and terms proportional to the coupling are kept small by the tadpole factor . by considering the ratio of the local vector current @xmath11 to the conserved vector current @xmath12
, we can compute the renormalization factor , @xmath13 we will consider a variety of choices for the initial and final states in this ratio , and the differences in the resulting values for @xmath8 will reflect the finite lattice spacing effects .
we have performed calculations at two values of @xmath14 with the improved action which correspond to lattice spacings of 0.4fm and 0.27fm as derived from the string tension . for comparison
, we have also used the wilson action to calculate at the same lattice spacings . for the gauge fields we used pseudo - heatbath updating with periodic boundary conditions ;
the first 4000 sweeps were discarded and then 250 sweeps ( 200 sweeps ) were discarded between each pair of improved ( wilson ) configurations that was kept . a stabilized biconjugate gradient algorithm was used for the fermion matrix inversion at three choices of the hopping parameter . due to the modest number of lattice sites ,
dirichlet boundary conditions were used for fermions propagating in the time direction but periodic boundaries were used in all spacial directions .
the source was placed two timesteps away from the boundary .
75mm@ccccc ' '' '' lattice & @xmath15 & @xmath14 & @xmath16[fm ] & @xmath17 + + @xmath18@xmath19 & 100 & 6.25 & 0.4 & 0.162 , 0.168 , 0.174 + @xmath20@xmath21 & 50 & 6.8 & 0.27 & 0.150 , 0.154 , 0.158 + + @xmath18@xmath19 & 75 & 4.5 & 0.4 & 0.189 , 0.201 , 0.213 + @xmath20@xmath21 & 30 & 5.5 & 0.27 & 0.164 , 0.172 , 0.180 +
figure [ fig : totz ] shows our determination of @xmath8 from four separate choices of the states in eq .
( [ ratio ] ) , @xmath22 the extrapolation to the chiral limit was done linearly in @xmath23 .
the results of wilson calculations by other groups are also shown .
notice that for both actions , @xmath8 at fixed @xmath24 is approximately independent of the amplitude used , except for the decay amplitude which predicts a smaller value for @xmath8 . for fixed lattice spacing ,
the improved action does give improved values of @xmath8 ; in fact the improved action matches the wilson predictions for @xmath8 when the lattice spacing is larger by a factor of about three .
figure [ fig : sinz ] shows that the dependence on @xmath25 is also quite similar for the two actions .
figure [ fig : fvnoz ] shows the _ lattice _ decay constant ( i.e. without @xmath8 ) .
again , we find that the improved action does give improved results when compared to wilson calculations at the same lattice spacing , and that the improved prediction is similar to the wilson prediction at 1/3 of the lattice spacing .
our best determination of the decay constant comes from the product of @xmath8 and @xmath26 , and this is shown in figure [ fig : fv ] .
the cluster of data points which lie above the experimental result rely on tadpole - improved perturbation theory to estimate @xmath8 .
notice that the completely nonperturbative determinations of @xmath27 are quite flat over a substantial range of lattice spacing , and that for both the wilson and improved actions the result is near experiment . in summary , we have found that the @xmath1 classical and tadpole improvements of our action manifest themselves in vector current matrix elements in essentially the same way as in masses@xcite .
improved action results for @xmath27 and @xmath8 are consistent with wilson action computations done at a lattice spacing about 1/3 the size .
this work was supported in part by the natural sciences and engineering research council of canada .
9 g. p. lepage and p. b. mackenzie , phys .
d48 ( 1993 ) 2250 .
h. r. fiebig and r. m. woloshyn , hep - lat/ 9603001 and these proceedings .
f. x. lee and d. b. leinweber , hep - lat/ 9606005 , to appear in the proceedings of panic 96 , world scientific .
m. lscher and p. weisz , comm .
97 ( 1985 ) 59 . h. hamber and c. m. wu , phys . lett .
b133 ( 1983 ) 351 ; w. wetzel , phys .
b136 ( 1984 ) 407 ; t. eguchi and n. kawamoto , nucl .
b237 ( 1984 ) 609 .
m. lscher and p. weisz , nucl .
b240 ( 1984 ) 349 . g. martinelli and c. t. sachrajda , nucl . phys . b316 ( 1989 ) 355 .
l. maiani and g. martinelli , phys .
b178 ( 1986 ) 265 .
y. iwasaki et .
al . , phys .
d53 ( 1996 ) 6443 . f. butler et .
al . , nucl .
b421 ( 1994 ) 217 . t. bhattacharya and r. gupta , phys .
d54 ( 1996 ) 1155 . | the @xmath0 meson decay constant and the associated renormalization factor are computed in the quenched approximation on coarse lattices using a tadpole - improved action which is corrected at the classical level to @xmath1 .
the improvement is displayed by comparing to wilson action calculations .
-1 cm |
the short period intermediate polar ( ip ) ex hydrae , belongs to the subclass of cataclysmic variables in which the strength of the magnetic field of the white dwarf ( wd ) is not strong enough to achieve synchronization between the rotation of the primary and the orbital period of the binary ( see * ? ? ? * and references therein ) .
the wd accretes from a surrounding disc or ring into the magnetic poles , producing a variety of observable phenomena ( e.g. * ? ? ?
* ; * ? ? ?
it was first identified by @xcite as an eclipsing system with an orbital period @xmath14 min , a disc inclination @xmath15 and a second prominent period @xmath16 min due to the rotation of the wd @xcite . the photometric visual light curve in ex hya is mostly dominated by the spin period which produces a sine - like modulation with an amplitude between @xmath17mag .
there are also occasional enhanced maxima and persistent narrow eclipses clearly associated with the orbital period cycle @xcite .
earlier determinations of the radial velocity semi amplitude of the primary , @xmath6 , have been obtained in the optical by several authors @xcite .
they estimated the masses of the binary from their derived @xmath6 values and the mass of the secondary from different mass - radius calibrations @xcite .
a first determination of @xmath6 from the fuv was obtained by @xcite .
high resolution results have been obtained from the euv and x - rays @xcite .
a first order estimation of the radial velocity semi - amplitude of the secondary , @xmath18 , was derived by @xcite using a cross - correlation technique and later by @xcite from and absorption lines .
mass determinations for the primary star have also been obtained by @xcite and @xcite from the x - ray spectra , as well from indirect methods by @xcite .
a wide variety of mass values for the binary components has been obtained from these publications .
its large @xmath19 ratio of @xmath20 places this system out of the usual spin equilibrium rotation value of @xmath21 @xcite .
if this is the case , the magnetosphere must fill the roche lobe of the primary star and there should be , at most , an accretion ring or no disc at all ( e.g. * ? ? ?
* ; * ? ? ?
. however , as pointed out by @xcite , there are several reasons to suggest that ex hya has a small magnetosphere far from equilibrium , among them are the eclipse timings of the partial x - ray eclipse which suggest a magnetospheric radius of only @xmath3 4 times the white - dwarf radius ; the lack of polarisation ; the long - term secular decrease and some of the radial velocity studies , which show the presence of an emission component that is modulated with the orbital period .
these controversies extend to the size of the inner disc .
the previous @xmath6 results have a large range of values .
this may be due to the variety of methods used to determine this value , or due to the use of a combination of emission lines , or obtained with very different spectral resolutions .
since we have obtained high - resolution spectroscopy in the red and done simultaneous visual photometry , we believe that our data could help to unravel the reason of these different results .
in particular , our analysis of h@xmath0 , includes a main disc as well as a broader accretion component .
ccd photometry with the johnson v filter was obtained in 2008 january 1013 , using the 1.5-m telescope with the marconi ccd for the first two nights and the 0.84-m for the rest of the nights with the thomson detector ( see table [ tab : logpho ] ) .
all ccd images were processed using the iraf package .
images were bias subtracted and flat field corrected .
a nearby comparison star with @xmath22 mag was used to determine the v magnitude of ex hya .
.log of photometric observations [ cols="^,^,^,^,^,^ " , ] our results from the individual doppler tomograms are similar to the overall tomogram shown in figure [ fig : streamall ] .
however , we detect substantial variations , particularly on the shape of the strong hotspot . in most nights
it appears well constrained and near the @xmath23 point , while in other cycles it appears highly smeared .
this smeared feature , or stream , could be correlated to a different combination of spin - orbital cycles .
further discussion can be found in section [ sec : discus ] .
the trail and reconstructed spectra in figure [ fig : trailspec ] are dominated by the strong hotspot and the stream .
the double - peaked accretion disc appears sometimes in the background with a peak - to - peak separation of @xmath31000 km s@xmath4 .
the contribution from the hotspot alone is clear in the narrow s - wave in the first and fourth nights and it shows a variation with an amplitude of @xmath3500 km s@xmath4 .
there seems to be an eclipse of the s - wave near phase zero in most of the trail spectra , nevertheless in trail 3 and 4 the eclipse appears to affect the red - shifted wing , while in trail 2 and 5 there is almost no eclipse at all .
furthermore , there is a bright spot in trail 2 near the s - wave . the stream is clearly shown in nights two and three , where we see a broader s - wave .
this stream is particularly strong , both in tomogram 5 and also in its corresponding trail spectra , where there is also a prominent contribution at phases 0.1 to 0.3 .
there is a red - shifted emission at phase 0.6 in trail 1 to 4 , extending from a slightly greater velocity amplitude than the hotspot .
this emission does not follow the double - peaked doppler shift and it encounters the red wing of the hotspot at phase 1.0
. a blue - shifted counterpart , occurring shorter in phase , is observed around phase 0.3 in trail 1 and 5 , and around phase 0.4 in trail 4 .
these features , combined with our radial velocity results , will be discussed in the next section .
ex hya is a well studied system , both photometrically and spectroscopically .
however , its spin to orbital period ratio , close to 2:3 , complicates its analysis and creates controversies , leaving still some unsolved problems . among them
is the existence or not of an accretion disc , and consequently the question of the size of the inner disc and the magnitude of its magnetosphere @xcite .
another remaining problem is the constancy of the orbital period .
while we have shown here that the orbital period remains constant , we can not rule out a possible small sinusoidal oscillation ( as far as our own observations ) . however the spin period is definitely decreasing at a slow rate @xcite . another important problem is the accurate determination of @xmath6 and the mass of the wd ( even in this eclipsing system ) , as we discussed in section [ sec : spec ] .
with respect to the problem of the radial velocity semi - amplitudes of the accretion disc ( and presumably of the primary star ) , we have pointed out that the published optical data have a large scatter ( see table [ tab : k1lit ] ) .
these works have been performed in the blue part of the visible spectra , while our observations are centred around a strong h@xmath0 emission line .
it is plausible that the disc component is more dominant in h@xmath0 than in higher order balmer lines and hence , the @xmath6 result would give a more accurate determination .
the similarity of the results in the radial velocity curves ( see section [ sec : spec ] ) , obtained from medium and high velocity regions have a deep implication : the high velocity component that we have analysed is moving in the orbital plane and therefore should be interpreted as an inner disc region .
in fact , our analysis indicate a robust result , both from the main disc component as well as from the broad wings .
furthermore , our @xmath5kms@xmath4value is in accord with the uv @xcite and x - ray @xcite results .
still , the observed rm - type effect remains a problem and defies easy explanation .
if this is the selective occultation of a rotating body , the simple explanation would be that the secondary star is partially occulting a symmetric disc . before ( but close to ) phase zero ,
the secondary star will ( partially ) occult first the blue shifted parts of the disc , while shortly afterwards , it will mask the receding parts of the disc . as a result
, the radial velocity distortions should be centred around phase zero , and this should be observed in both , the main disc and the inner disc .
we observe , however , that the asymmetry in both cases , is centred around phase 0.9 .
this implies that there is an asymmetric component present , and there is only one asymmetric component in the disc that could account for this shift , namely , the hot spot .
we have calculated that the phase offset @xmath24 between eclipses and inferior conjunction is around 0.1 ( see table [ tab : orbparam2 ] ) .
although we have avoided measuring the hot spot velocity component , it is possible that the hot spot itself is physically occulting the high velocities behind it .
therefore , we propose that the rm - type effect is in this case a combination of the occultation of the disc velocity regions by both the hot spot and the secondary star . this explanation is supported by the position of the spot shown in the tomography and by the fact that the rm - type effect moves closer to phase zero for the broad wings ( see figure [ fig : velcur2 ] ) , suggesting that the effect of the occultation by the hot spot is smaller for higher velocity regions .
in other words , the hot spot eclipse is a grazing eclipse and does not affect the central parts of the disc , nor the white dwarf itself . with respect to the accurate determination of the masses of the binary , we believe that our careful determination of @xmath6 by two different methods , consistent with uv and x - ray measurements , coupled with the accurate determination of @xmath18 by @xcite , enable us to derive reliable masses ( see section [ sec : baspar ] ) .
the obtained values are consistent with a short period system with a primary mass with a mean value of 0.8 m@xmath10 ( e.g. @xcite ) and a secondary star with enough mass to produce nuclear reactions , still approaching the minimum orbital period @xcite .
a long standing discrepancy with the mass of the primary obtained indirectly by x - rays , has also been resolved recently by @xcite .
the results of the doppler tomography are quite revealing .
although the overall tomogram , shown in fig [ fig : streamall ] , does not reveal the weak broad component , due to the dilution effect in a standard projection , mentioned in section [ sec : dopmap ] , the trail spectra does show this high velocity component ( up to @xmath32000 kms@xmath4 ) .
the individual tomograms , show that there are fast changes in the hot spot or stream from cycle to cycle , and particularly from night to night , revealing different combination of spin - orbital cycles . to illustrate the effects of the spin - orbital configuration in the tomography , as well as in other related aspects discussed below , we present in figure [ fig : diagram ] , three diagrams of the inner disc and the magnetosphere in a classical configuration ( e.g. @xcite ) .
( we must clarify at this point that these schematics are a visual help only and we do not intend to quantify any physical parameter from it ) .
spin phases 0.0 and 0.5 are defined at the time when the visible magnetic pole points away and towards the observer respectively .
each one has also been labelled for orbital phases 0.0 ( when the secondary star is in front of the observer ) and 0.5 ( when the secondary is behind the disc ) . the hot spot and stream episodes appear connected with the spin - orbital cycle combination .
two examples are selected , which have the desired photometric combination of spin and orbital phases ( see figure [ fig : vphot ] ) .
an episode with a more pronounced stream would correspond , as seen mainly in tomogram 4 , when the material is smeared because the magnetic pole points toward the secondary star ( _ lower diagram _ , @xmath25 , @xmath26 ) ; a less pronounced stream , e.g. tomogram 1 , would produce a more compact spot , when the pole points away from the observer but the secondary star is again at inferior conjunction ( _ top diagram _ , @xmath27 , @xmath26 ) .
all other tomograms show , either a contained hot spot or a hot spot and a stream , but in these cases , although the spin and orbital phases are at some point in conjunction or opposition , the photometry has a complex behaviour , which makes the interpretation of these streams or compact spots more difficult
. this would be the case of tomogram 3 , a stream seems to dominate due to the high photometric activity near spin cycle 42 .
it is clear that these stream or compact spots do not affect our radial velocity results , as the hot spot and or stream , have been avoided by our double gaussian of the main disc component and single gaussian analysis of the broad component .
although the h@xmath0 radial velocity curves depicted in figures [ fig : velcur1 ] and [ fig : velcur2 ] have a clear orbital modulation , the spin and orbital modulations have a mixed behaviour .
for example , the ew and more clearly , the relative flux , is modulated mainly with the spin cycle , but this modulation also depends on the orbital phase . as an example , we show in figure [ fig : mod_all ] ( lower panel ) that the maximum flux occurs when the magnetic upper pole of the wd points away from the observer , at spin phase 1.0 and the secondary is behind the disc , while a minimum is seen at the same phase when the secondary is in front ( figure [ fig : mwings ] , top panel ) .
our analysis of the broad component presented in sections [ sec : radvel2 ] and [ sec : swidth ] shows that the radial velocity behaviour of the inner region is very similar to that of the disc component .
this suggests that the observed broad wings arise from material predominantly coming from the inner disc region , still in the orbital plane , although there may be also some small contributions from the accretion curtain . as we described earlier ,
the width of the broad component is modulated with the spin period ( figure [ fig : mwings ] , top panel ) .
we interpret the maximum broadening around phase zero as a preferential fan - like view of the inner accretion curtain , when the magnetic upper pole points away from the observer .
the grazing eclipse of the secondary star would not affect this broadening ( see figure [ fig : diagram ] , top diagram ) as it will cover the lower magnetic pole only . however , the behavior of the relative flux of the broad line is not like the one we measure for the whole line ( see figure [ fig : mod_all ] , lower panel ) . in section [ sec : swidth ]
we explained that , although the flux is modulated by the spin - cycle , there is a clear correlation when the spin and orbital phases coincide . in the case of the broad wings ,
there is no correlation with the spin cycle ( not shown in the paper ) , but there is an apparent double modulation with the orbital period .
we must point out again , that the individual nights show different results , again depending on the correlation between the spin and the orbital phases , shown in the right panel of figure [ fig : mwings ] .
the double modulation is clearly seen in the first night , where the spin and the orbital phases do not coincide at phase zero .
however , in night two , the flux is dominated by an eclipse at phase zero , when both spin and orbital phases coincide .
although there seems to be a maximum at phase 0.75 , we do not have sufficient orbital coverage on that night to claim the double modulation .
where as in night three , which covers two spin cycles , there is a single modulation with a maximum at phase 0.25 and a shallow eclipse at phase zero . in night four , which also covers two spin cycles
, there is only a large scatter in flux values .
we also observe that the relative flux of the broad component is more sensitive to this spin - orbital correlation , than the flux observed for the whole line .
we are uncertain , therefore , that the double modulation in flux is a real and stable feature .
we believe that more observations are needed to confirm it .
our radial velocity study agrees with the main results by @xcite , yielding similar masses for the binary .
but their range in @xmath6 values implies a primary mass between 0.62 and 0.99 m@xmath10 with a most likely value of 0.78 m@xmath10 .
the mass of the secondary star is obtained indirectly using an empirical zams calibration by @xcite . the fact that the masses are in agreement is due to the use of our smaller @xmath6 and to the adopted value for @xmath18 from @xcite . nevertheless , the mass result of the primary in both cases is close to 0.8 m@xmath10 , and it is also encouraging that the new cooling flow models from the x - ray emission , now agrees with this mass @xcite .
the eclipses seen in the trail spectra favour the interpretation of the s - wave component as a bright emission region near the outer edge of the circulating disc as stated by @xcite .
the s - wave asymmetry , noted by these authors , is seen , but only in our trail spectra 1 and 3 with a more rapid transition from blue to red .
the rest show a more sinusoidal behaviour as seen in trail 4 , 5 and 6 .
trail 2 has a gap between phases 0.1 and 0.3 making difficult to tell .
we should note that our orbital phases are 0.1 ahead of those of @xcite due to the use of our new ephemeris .
this creates only a shift and have no influence on the asymmetries or lack of them .
the tail - like features below the s - wave , seen in most of our trail spectra extending to slightly higher velocities , are also observed by @xcite .
they interpret these as a high velocity component caused by an overflow stream , reaching @xmath28kms@xmath4 , although , in our case , they only reach velocities up to 600kms@xmath4 . in addition , our reconstructed spectra 5 , suggest that these tails may form a second s - wave , shifted 0.5 in phase with the main s - wave , in contrast with a 0.2 shift observed in the reconstructed spectra by @xcite .
the reconstructed spectra 3 and 4 show a broader s - wave , in accordance with this stream , but still with velocities no greater than 600kms@xmath4 .
@xcite estimates a magnetospheric radius of about 4 white - dwarf radii ( @xmath29 ) from the eclipse timing analysis which locates the centroid of the spin - varying emission @xcite .
we have to point out that the spin - varying emission does not hold in the long term ( figure [ fig : omc ] , _ bottom _ ) .
we believe that this effect and the possible sinusoidal variations with the orbital period might be due to a changing geometry of the system concerning the material locked by the magnetosphere and to the position of the hot spot .
nevertheless , our h@xmath0 emission is strong in a region up to @xmath31000kms@xmath4 , and this would result in a minimum accretion disc radius of @xmath30 , while the circularisation radius for a system with @xmath31 ( * ? ? ?
4 ) turns out to be @xmath32 .
however , if our high velocity component comes from a region with keplerian velocity ( i.e. an inner disc component still in the orbital plane ) , then the minimum disc radius would extend only to @xmath13 .
this is in accordance with the fact that ex hya shows no polarisation , which is unlikely if the magnetosphere is to dominate the white - dwarf s roche lobe @xcite .
thus , our results point ex hya to have a small magnetosphere and being far from equilibrium .
although we have here claimed that the broad component comes from a disc - like structure , why does the emission line profile have an inflection point around 1000 kms@xmath4 ? and why is this component modulated mainly by the orbital modulation and not by the spin modulation as in the case of the whole line ? does this inflection point affects our result on our calculation of the inner radius of the disk ?
these are questions that need to be raised for future studies of ex hya ; further observations are suggested to look closely at these points .
1 . our photometric observations of ex hya show a strong modulation with the 67 min spin period and 0.4mag narrow eclipses associated with the 98 min orbital period .
we have computed new ephemeris for ex hya : @xmath33 3 .
the radial velocity analysis yields a semi - amplitude @xmath34 km s@xmath4 , both from the disc and broad components which , combined with the @xmath18 value from @xcite and an inclination angle of @xmath35 , agreed by most authors , gives a determination of @xmath9 m@xmath10 , @xmath11 m@xmath10 and @xmath12 r@xmath10 .
the equivalent width of the whole line is slightly modulated with the spin period as found by previous authors .
furthermore , we find that its relative flux is strongly modulated with the spin period , but their maxima and minima , depend on the alignment of the spin pole and the inferior conjunction of the secondary star
the gaussian sigma of the fit of the broad component is also clearly modulated with the spin period , even more clearly than the equivalent width of the whole emission line .
there is a maximum at phase 1 and a minimum at phase 0.5 .
we interpret this as a fan - like view of the inner accretion curtain , when the magnetic upper pole points away from the observer .
however , the relative flux of this broad component shows an apparent double modulation with the orbital period . due to the dissimilarities in the individual nights , we are uncertain that this modulation is a real and stable feature and believe that more observations are needed to confirm it . 6 .
we propose that the rm - type effect is the result of the combination of the occultation of the disc velocity regions by both the hot spot and the secondary star ; an explanation which could account for the varying phase shift of the distortion .
the doppler tomography reveals a well formed disc and a strong hot - spot , at times smeared .
the changes in shape of the hot - spot could be due to a different combination of spin - orbital cycles .
the observational analysis we have made in this paper , including the radial velocity study , the simultaneous photometry and the doppler tomography , indicate that ex hya has a well formed disc and an inner disc region , both which shows the same radial velocity behaviour .
there is also a an indication of a small contribution from the accretion curtain . from the velocity of the broad wings
, we estimate that the magnetosphere should extend only to about @xmath13 .
the authors are indebted from dgapa ( universidad nacional autnoma de mxico ) support , papiit projects in111713 and in122409 .
this article is based upon observations collected at the observatorio astronmico nacional at san pedro mrtir , b.c .
we also thank the staff of the observatory for their invaluable help and to juan carlos yustis for producing the inner disc diagram .
jvhs acknowledges support via studentships from conacyt ( mxico ) and the university of southampton .
we would like to thank the anonymous referee for the prompt response and excellent feedback and to rafael costero for his invaluable insight into the physics of this system .
99 belle , k. e. , howell , s. b. , sirk , m. m. & huber , m. e. 2002 , , 577 , 370 belle , k. e. , howell , s. , sion , e. , long , k. s. , szkody , p. 2003
, , 587 , 373 beuermann , k. , harrison , t. e. , mcarthur , b. e. , benedict , g. f. , & gnsicke , b. t. 2003 , , 412 , 821 beuermann , k. & reinsch , k. 2008 , , 480,199 beuermann , k. baraffe , i. , kolb , u. & weichhold , m. 1998 , , 339 , 518 bond , i.a . &
freeth , r.v .
1988 , , 232 , 753 breysacher , j. & vogt , n. 1980 , 87 , 349 cowley , a. p. , hutchings , j. b. & crampton , d. 1981 , , 246 , 489 cropper , m. , ramsay , g. , & wu , k. 1998 , , 293 , 222 echevarra , j. 1983 , revmexaa , 8 , 109 eisenbart , s. , beuermann , k. , reinsch , k. & gnsicke , b. t. 2002 , , 382 , 984 fujimoto , r. , & ishida , m. 1997 , , 474 , 774 gilliland , r. l. 1982 , , 258 , 576 hellier , c. , mason , k. o. , rosen , s. r. , & cordova , f. a. 1987 , , 228 , 463 hellier , c. 2014 .
epj web of conferences , 64 07001 hellier , c. & sproats , l. n. 1992 , ibvs , 3724 , 1 hill , k. m. , sharma , d.p . & watson , r.d .
1985 , _ poster paper , iau symp .
no 228 _ , christchurch , new zeland howell , s. b. , nelson , l. a. & rappaport , s. 2001 , , 510 , 897 hoogerwerf , r. , brickhouse , n. s. , & mauche , c. w. 2004 , , 610 , 411 jablonski , f. & busko , i.c .
1985 , , 214 , 219 king , a. r. & lasota , j .-
1991 , , 378 , 674 king , a. r. & wynn d. w. 1999 , , 310 , 203 kotze , e.j . , potter , s.b . & mcbride , v.a . 2015 , , 579 , 77 knigge , c. 2006 , , 373 , 484 kolb , u .. & baraffe , i. 1999 , , 309 , 1034 kraft , r. p. 1962
, , 135 , 408 kruszewski , a. , mewe , r.,heise , j. et al .
1981 , space sci .
, 30 , 221 kaitchuck , r. h. , hantzios , p. a. , kakaletris , p. , honeycutt , r. k. , schlegel , e. m. 1986 , , 317 , 765 luna , g. j. m. , raymond , j.c . , brickhouse , n.s , mauche , c.w . & suleimanov , v. 2015 , , 578 , 15 marsh , t. r. , & horne , k. 1988 , , 235 , 269 mauche , c. w. 1999 , , 520 , 822 mauche , c. w. , brickhouse , n. s. , hoogerwerf , r. , luna , g. j. m. , mukai , k. & sterken , c. 2009 , ibvs , 5876 mclaughlin , d.b .
, , 60 22 mhlahlo , n. , buckley , d. a. h.,dhillon , v. , potter , s. b. , warner & b. , woudt , p. 2007
, , 378 , 211 mumford , g.s .
1967 , , 190 , 637 norton , a. , wynn , a. & sumerscales , c. 2004 , , 614 , 349 patterson , j. 1984 , , 54 , 443 patterson , j. 2011 , , 411 , 2695 robinson , e. l. 1976 , in _ ann .
_ , 14 , 119 rosen , s. r. , mason , k. o. & mukai , k. 1991 , , 249 , 417 rossiter , r.a . , , 60 15 sterken , c. , vogt , n. , freeth , r. , kennedy , h. d. , page , a. a. & walker , w. s. g. 1983 , , 118 , 331 shafter , a. w. , szkody , p. & thorstensen , j. r. 1986 , , 308 , 765 siegel , n. , reinsch , k. , beuermann , k. , van der woerd , h. & wolff , e. 1989 , , 225 , 97 vogt , n. , krzeminski , w. , & sterken , c. 1980 , , 85 , 106 vande putte , d. , smith , r. c. , hawkins , n. a. & martin , j. s. , 2003 , 342 , 151 warner , b. 1976,in _ structure and evolution of close binaries systems _ , iau symp . 73 .
eggleton et al .
dordrecht , holland , p85 .
warner , b. 1995 , cataclysmic variable stars , cambridge university press , cambridge . | a study on the intermediate polar ex hya is presented , based on simultaneous photometry and high dispersion spectroscopic observations , during four consecutive nights .
the strong photometric modulation related to with the 67-min spin period of the primary star is clearly present , as well as the narrow eclipses associated to the orbital modulation . since our eclipse timings have been obtained almost 91,000 cycles since the last reported observations , we present new linear ephemeris , although we can not rule out a sinusoidal variation suggested by previous authors .
the system shows double - peaked h@xmath0 , h@xmath1 and @xmath25876 emission lines , with almost no other lines present . as h@xmath0 is the only line with enough s / n ratio in our observations , we have concentrated our efforts in its study , in order to obtain a reliable radial velocity semi amplitude . from the profile of this line
, we find two important components ; one with a steep rise and velocities not larger than @xmath31000kms@xmath4and another broader component extending up to @xmath32000kms@xmath4 , which we interpret as coming mainly from the inner disc . a strong and variable hotspot is found and a stream - like structure is seen at times .
we show that the best solution correspond to @xmath5kms@xmath4 from h@xmath0 , from the two emission components , which are both in phase with the orbital modulation .
we remark on a peculiar effect in the radial velocity curve around phase zero , which could be interpreted as a rositter - maclaughlin - like effect , which has been taken into account before deriving @xmath6 .
this value is compatible with the values found in high - resolution both in the ultraviolet and x - ray . using the published inclination angle of @xmath7 and semi - amplitude @xmath8kms@xmath4
, we find : @xmath9 m@xmath10 , @xmath11 m@xmath10 and @xmath12 r@xmath10 .
doppler tomography has been applied , to construct six doppler tomograms for single orbital cycles spanning the four days of observations to support our conclusions .
our results indicate that ex hya has a well formed disc and that the magnetosphere should extend only to about @xmath13 .
stars : novae , cataclysmic variables techniques : radial velocities eclipses ephemerides |
epitaxial combination of superconducting films and ferroelectric layers is promising for construction of effective microwave devices .
especially advantageous is the application of small - loss high - temperature cuprate superconductors ( ybco)@xcite and perovskite ferroelectric alloys like ba@xmath0sr@xmath1tio@xmath2 ( bst ) and pb(zr@xmath0ti@xmath1)o@xmath2 ( pzt ) , in heterogeneous miltilayers .
as both of these compounds have similar perovskite - type crystal structure with small lattice mismatch , the well oriented bst and pzt thin layers can be grown on the surfaces of ybco@xcite .
in addition , the role of strains and off - stoichiometry structural defects at the bst(pzt)/ybco - interfaces can be minimized by the use of growing methods such as laser ablation for fabrication of the miltilayered structures@xcite . since the 1960s , the strong interest in developing electric field - effect devices based on ferroelectrics has been stimulated by observations of shifts of superconducting transition temperature under a switch of the spontaneous polarization . superconducting field - effect transistors ( sufets ) can not be realized with conventional superconductors since the large coherence lengths and high carrier density of the latter result in negligibly small effect of the polarization field @xcite .
in contrast , the carrier density and coherence lengths of high - t@xmath3 cuprates ( hts ) which are two orders of magnitude lower than those in conventional superconductors@xcite , have made their use in sufets very attractive @xcite .
for example , it was demonstrated in ref .
, that the critical temperature of ybco films could be shifted by about 6 k through the polarization in the batio@xmath2 substrate . as was assumed in @xcite , the increase of @xmath4 is connected with the formation of the accumulation layers ( with charge carrier density enhanced due to the polarization directed toward the interface ) , whereas the polarization in the opposite direction induces a depletion layer with higher resistance and lower @xmath4 .
besides critical temperature , ferroelectric polarization affects transport properties of sufets leading to a change in resistance of 9 - 25 @xmath5 for pzt - ybco and pzt - ndba@xmath6cu@xmath2o@xmath7(nbco ) heterostructures @xcite and showing polarization - induced memory effects in resistance and current , which suggests their possible storage applications@xcite . in the case when the polarization is parallel to the fe - s interface , the use of the zero-@xmath8 boundary conditions ( @xmath9 at the electrode ) results in a weak effect of the fe - layers scarcely affecting the behavior of the s - films . on the other hand , the influence of the fe - layer should be essentially strong when the fe - polarization vector @xmath10 is perpendicular to the fe - s interface .
due to the specific perovskite - like structure , the properties of the high - t@xmath3 materials are believed to be very close to the perovskite - type ferroelectrics .
this fact stimulated discussions about a coexistence of ferroelectricity and superconductivity and a possible ferroelectricity of hts - compounds@xcite .
as a consequence , the ferroelectric polarization has been suggested to be nonzero at the interface with the hts - layer with even the possibility of penetration inside the superconductor@xcite .
recent studies of ferroelectric - hts layered structures were focused on investigations of the role of free @xmath8-boundary conditions ( non - zero @xmath8 ) on dielectric properties of fe - layers@xcite .
however , little attention has been paid to the behavior of superconducting films as well as to the effect of the fe - s interface on ferroelectric structural transformations . in this work ,
we develop a phenomenological model based on the ginzburg - landau theory , for the fe - layers sandwiched between superconducting films . as compared to the microscopic approach proposed in ref .
where the s - layer was described as two s - planes with a possibility of a charge transfer between the planes and the interior of the layer , the proposed model allows one to study directly the behavior of the system for different thicknesses of the s - layer in the vicinity of the ferroelectric transition temperature in the fe - film , which is assumed to be close to the transition temperature of the superconductor .
we note that for hts - compounds the condition @xmath11 for the ginzburg - landau theory is valid over a much wider temperature range than in conventional superconductors @xcite .
we focus primarily on the case when the penetration depth of the polarization - induced electric field is larger than the superconducting coherence length ( @xmath12 ) , which is also typically realized in hts layers .
the paper is organized as follows . in section [ model ] , we introduce the ginzburg - landau energy functional for the fe - s heterostructure .
we derive the ginzburg - landau equations and elucidate the nature of the boundary conditions . in section [ phases ] , we analyze in details the results for the case when the bulk ferroelectric transition temperature lies below the temperature of the superconducting transition .
we discuss possible phases which can be stabilized in the system depending on the values of the temperature and the coupling energy .
corresponding phase diagrams are analyzed in section [ diagrams ] for different thicknesses of the fe- and s - layers .
we discuss here also the thicknesses - dependences of the ferroelectric transition temperature .
concluding remarks are presented in section [ conclusions ] .
we consider a system containing periodically alternating ferroelectric ( fe ) layer sandwiched between superconducting ( s ) films ( shown in fig .
[ fig0 ] ) infinitely extended in the @xmath13 and @xmath14 directions .
we assume here that the polarization @xmath15 is directed perpendicular to the fe - s interface .
the thicknesses of the fe- and s - layers are given by @xmath16 and @xmath17 respectively , and thus the total width of the supercell is : @xmath18 . generally , the whole sandwich structure can be described by the following ginzburg - landau functional @xmath19 the ferroelectric part of the free energy ( [ g_l ] ) is given by @xmath20,\ ] ] where @xmath21 is the inverse of the ferroelectric susceptibility and @xmath22 denotes the bulk curie temperature . for simplicity
, we do not include the sixth - order term into the @xmath8-expansion ( [ f_ferr ] ) , restricting our analysis to second - order phase transformations .
note that since the polarization depends on @xmath23 only , the last term in ( [ f_ferr ] ) simplifies to @xmath24 .
= 8.5 cm for the superconducting s - film without magnetic field the free energy can be written as @xmath25,\ ] ] where @xmath26 with @xmath4 denoting the bulk superconducting transition temperature , and @xmath27 , @xmath28 and @xmath29 are the mass , charge and the local density of the cooper pairs respectively . similarly to ( [ f_ferr ] ) we assume in ( [ f_s ] ) that @xmath30 depends on @xmath23 only , i.e. , @xmath31 . note also that without a magnetic field the phase @xmath32 , and the vector potential @xmath33 ( see the appendix ) .
the ginzburg - landau equations in this static case contain only the terms with @xmath34 and below for simplicity we use the notations @xmath35 .
the interface part of the free energy ( [ g_l ] ) describes the interaction between the spontaneous polarization and the superconducting charge near the interface . in order to derive @xmath36 , we consider first the interaction of @xmath8 with an electric field @xmath37 associated with the polarization distribution @xcite @xmath38 where @xmath39 is the electric displacement .
the field @xmath40 can be determined from the maxwell equation inside the fe - layer @xmath41 .
the integration in the vicinity of the contacts gives us near @xmath42 @xmath43 and near @xmath44 @xmath45 in contrast to the normal metal electrodes where @xmath9 and @xmath46 , we employ the following two concepts for the fe - s interface : ( i ) @xmath47 ( see the corresponding discussion in introduction ) ; ( ii ) an electric field can penetrate inside the superconductor ( @xmath48 ) and thus substantially influence the superconducting properties@xcite .
using these assumptions together with the boundary conditions for @xmath49 @xcite @xmath50 where @xmath51 denotes the charge density induced by the polarization at the fe - s contact and @xmath52 is the unit vector directed perpendicular to the surface of the ferroelectric into the s - layers , we obtain from ( [ ed_0 ] ) and ( [ ed_l1 ] ) @xmath53 to derive the expressions for @xmath54 penetrating inside the s - film at a distance of the thomas - fermi charge screening length @xmath55 , we consider the poisson s equation near the right ( @xmath56 ) and left ( @xmath57 ) fe - s contacts @xcite @xmath58 where @xmath59 , with the boundary conditions for the electrostatic potential @xmath60 @xmath61 here @xmath62 is larger for systems with the lower mobile charge carrier concentration @xmath63 , and @xmath64 is the local fermi energy .
the solutions obtained from ( [ poisson ] ) , ( [ pois_bound ] ) decay exponentially inside the right s - film ( @xmath65 ) @xmath66 and the left s - film ( @xmath67 ) @xmath68 the electric field penetration depth @xmath55 in hts compounds is shown to be in the range of about @xmath69 nm ( ref . ) , which is significantly larger than @xmath70 nm in the normal metal electrodes .
note that the total charge in the s - layer : @xmath71 includes the electron and background contributions : @xmath72 with the electron charge consisting of the normal and superconducting components : @xmath73 according to the two - fluid model of superconductivity @xcite . since we study in this work the superconducting contribution @xmath74 , we assume here that the induced surface density @xmath51 is described by a local increase / decrease of the superconducting charge density @xmath75 where @xmath76 is the coupling coefficient characterizing the change of the charge carrier density near the contacts due to the spontaneous polarization in the fe - layer .
from the expressions ( [ ed_b ] ) , ( [ sigma ] ) , and ( [ p_e_d ] ) we obtain @xmath77 , where the interaction energy inside the fe - layer @xmath78 the electrostatic interaction of the penetrating field @xmath54 given by ( [ e_pen_r])-([e_pen_l ] ) , with the superconducting charge @xmath79 near the contacts ( @xmath56 ) and ( @xmath80 ) is determined by @xmath81 at a small distance @xmath82 near the contact @xmath44 @xmath83 and @xmath42 @xmath84 and the expressions ( [ e_pen_r ] ) and ( [ e_pen_l ] ) for @xmath60 . as the result , we get @xmath85 where @xmath86 , @xmath87 and @xmath88 . the first term ( [ f_int ] )
renormalizes the coefficient @xmath89 in ( [ f_ferr ] ) and leads to a symmetrical reduction of the deviation of @xmath8 from its bulk value near the boundaries as was shown in ref . .
as a first step in the analysis of the role of the interface charge carriers , in this work we focus our attention on the contribution of the first two interface terms in the @xmath90 given by ( [ f_fe_s ] ) , neglecting the smaller higher order terms in the @xmath91-expansion and disregarding the well studied term ( [ f_int ] ) . it should be noted that the contribution ( [ f_fe_s ] ) can be derived from a microscopic model describing an ising - type fe - layer sandwiched between the s - planes with bcs - pairing@xcite . in the latter approach ,
the coupling energy @xmath92 characterizes the change of the one - electron energy @xmath93 in the s - plane due to the distortion of the nearest atomic group of the fe - layer with the coordinate @xmath94 ( the average distortion @xmath95 becomes nonzero in the ferroelectric state ) .
minimization of the total free energy @xmath96 where @xmath97 denotes the ( @xmath13 , @xmath14)-surface , by variation of the order parameters @xmath8 and @xmath98 yields the following two ginzburg - landau equations @xmath99 the extra surface terms appearing from the variation , give the set of the boundary conditions at the right contact @xmath100 @xmath101 and at the left contact @xmath102 ( @xmath103 ) @xmath104 setting @xmath105 in ( [ bz_l1 ] ) and ( [ bz_l ] ) results in ordinary free boundary conditions ( zero gradients of the order parameters ) at the fe - s contacts , which implies in fact the continuity of @xmath8 and @xmath98 near the boundaries
. however , a small finite coupling energy @xmath106 gives non - zero gradients of @xmath8 and @xmath98 @xmath107 where @xmath108 and @xmath109 denote the corresponding contact values .
consider the first expression ( [ cont ] ) for @xmath8 .
as the right hand side is always positive , we conclude that @xmath110 for @xmath111 this relation gives at the right contact @xmath44 an increase of @xmath8 due to the coupling with the surface charge , whereas at the left contact @xmath42 , @xmath8 decreases .
since for @xmath112 we have the similar behavior , we conclude that independent of the direction of the polarization @xmath113 let us analyze the second condition ( [ cont ] ) for @xmath98 which in distinction to the latter case depends on the direction of the polarization . for @xmath111 , the gradient @xmath114 and we obtain an increase of the electronic pair density at @xmath44 and a lower density at @xmath115 .
furthermore , a switching of the polarization ( @xmath112 ) results in @xmath116 and in the opposite behavior of the charge density at the contacts . below we present the numerical results for the problem ( [ g_lan_eq])-([bz_l ] ) in the case when @xmath117 , discussing in limiting cases the analytical solutions for @xmath8 and @xmath98 .
note that the case @xmath117 is realized in complex insulating perovskite alloys such as bst with high sr content .
the ferroelectric transition temperature in bst depends on the ba content @xmath13 , approaching @xmath118 for @xmath119 ( srtio@xmath2 is a paraelectric material with no ferroelectric phase transition ) and @xmath120 c for @xmath121 ( batio@xmath2 ) .
for instance , for @xmath122 ba , @xmath123 k @xcite which lies already in the range of superconducting transition temperatures of the hts compounds .
we now present a brief overview of the results obtained for our model ( [ g_lan_eq])-([bz_l ] ) for different temperatures and coupling @xmath91 .
[ fig1 ] shows spatial dependences of @xmath8 and @xmath98 normalized by the zero - temperature values @xmath124 and @xmath125 for the following set of the parameters : @xmath126 ( @xmath127 , @xmath128 ) and @xmath129 at a low temperature @xmath130 , while @xmath131 . here
@xmath132 is the zero - temperature superconducting coherence length and the correlation length for the fe - layer at @xmath133 is : @xmath134 .
note that taking into account the possible variation of the field penetration depth ( section [ model ] ) , we can estimate the range for the magnitude of the dimensionless coupling @xmath135 : from @xmath136 ( for conventional superconductors ) to @xmath137 ( for hts films ) .
for the films of moderate thickness ( @xmath138 and @xmath139 ) , one can observe a strong coupling effect demonstrated in fig .
[ fig1 ] . similarly to the theory for bulk ferroelectrics ,
the two possible directions of the polarization can be stabilized in the fe - layer . as was already discussed in section [ lan_ginz_eq ] ( equations ( [ cont ] ) and ( [ cont_1 ] ) ) , @xmath8 increases at the right contact , @xmath44 , and lowers at the left one due to the coupling with the s - electrons ( fig . [ fig1](a ) ) .
the important effect arising from this , is the appearance of two stable solutions for @xmath98 in the adjacent s - layer ( fig .
[ fig1](b ) ) .
the first of them ( with @xmath140 , @xmath141 ) corresponds to the positive branch of @xmath8 whereas the second one ( @xmath142 ) appears due to the coupling with the negative @xmath8 .
note that the regions with the lower density @xmath143 ( the depletion layer ) and the accumulation charge are located near the contacts ( @xmath144 ) and ( @xmath145 ) with @xmath146 , whereas inside the thick s - layer ( fig .
[ fig1](b ) , inset ) , @xmath98 approaches its bulk value @xmath147 .
= 8.5 cm = 8.5 cm similar interface effects with an increase of the fe - layer thickness are demonstrated in the insets of fig .
[ fig1](a ) , where the regions of the deviations of @xmath8 from its bulk value @xmath148 are located within a distance @xmath149 from the interfaces .
the whole physical behavior becomes highly nontrivial for @xmath150 ( @xmath151 ) in the temperature range @xmath152 .
in contrast to the bulk , where @xmath153 for @xmath150 , the coupling with the interface charge leads to rather unusual behavior for @xmath8 and @xmath98 as shown in fig .
[ fig2 ] at @xmath154 .
the polarization in this case is enhanced , and for @xmath155 and small @xmath156 is given by @xmath157 where the superconducting pair density at @xmath103 @xmath158+\mathcal{o}(\gamma^4).\ ] ] as follows from ( [ p_enh ] ) , in this state a non - zero @xmath8 exists only due to the superconducting electrons , because setting @xmath105 or @xmath159 immediately yields @xmath153 . for the thin fe - films with @xmath160 or close to the bulk transition temperature ( where @xmath161 ) , the effect of the contacts is particularly pronounced and the polarization is linear in @xmath23 @xmath162 however , as the fe - layer becomes thicker ( @xmath163 ) , the polarization deviates from zero only near the interfaces , whereas inside the layer @xmath164 .
for example , for @xmath165 @xmath166 and decays exponentially as one moves towards the interior of the layer . near the right contact @xmath167 , the behavior is similar @xmath168 note that the polarization is antisymmetric with respect to the center of the fe - layer @xmath169 : @xmath170 , forming two @xmath8-domains : with a negative @xmath8 for @xmath171 , and a positive @xmath8 for @xmath172 , as shown in fig .
[ fig2](a ) .
the center of the layer , where @xmath173 , corresponds to the domain wall .
= 8.5 cm = 8.5 cm consider now the behavior of the s - layer .
for thin s - films ( @xmath174 , @xmath175 ) in the vicinity of the @xmath22 ( @xmath155 ) , and for a weak coupling @xmath156 , the superconducting order parameter is given by @xmath176+\mathcal{o}(\gamma^4).\ ] ] in contrast to the antisymmetric behavior of @xmath8 in the domain - type phase , @xmath177 is symmetric about the center of the s - layer @xmath178 : @xmath179 as can be seen in fig . [
fig2](b ) .
we note that for the case of thin s- or fe - films ( @xmath180 or @xmath181 ) the effect of fluctuations is expected to be essentially important .
this can be studied by including the additional higher - order terms into the free - energy functional ( [ f_ferr])-([f_s ] ) that will be a subject of separate work . from ( [ psi_thin ] )
it follows immediately that as @xmath182 or @xmath183 , @xmath177 approaches its bulk value @xmath184 .
moreover , we obtain the bulk behavior in ( [ psi_thin ] ) also when the adjacent fe - film becomes thin , so that @xmath185 . similarly to the fe - layer , in a thick s - layer the influence of the interfaces is remarkable only near the contacts ( fig . [ fig2](b ) , inset ) . to gain a better understanding of the nature of the domain phase , we determine its free energy which can be obtained by the substitution of @xmath8 and @xmath98 from ( [ p_enh ] ) and ( [ psi_thin ] ) into ( [ f_ferr ] ) , ( [ f_s ] ) and ( [ f_fe_s ] ) @xmath186 note that since for the free ( non - coupled ) fe - layer @xmath187 for @xmath150 , the total free energy for the non - coupled fe+s multilayers for @xmath155 is given by @xmath188 therefore , the domain state always has an energy gain due to a negative difference between the energies in these two phases @xmath189 the latter result supports our conclusion that a non - zero coupling with @xmath190 is the main reason for the stabilization of the domain state .
we note also that as @xmath182 , we obtain that the energy gain @xmath191 , the polarization @xmath164 , and hence the domain phase transforms in fact into an ordinary paraelectric state : @xmath192 thus in the following analysis we define @xmath193 as the temperature of the transition between the ordered fe+sc phase with the bulk - type @xmath194 , and the domain - type phase , which occurs on heating . as results from the previous section ,
generally a weak coupling @xmath195 at the contacts produces symmetric domain phase with a non - zero polarization for @xmath155 .
we demonstrate how this behavior changes as the coupling becomes stronger , i.e. , with the increasing @xmath135 . in fig .
[ fig3 ] , we plot the order parameters vs @xmath135 for layers of thicknesses @xmath196 and @xmath197 . at lower @xmath198 ( @xmath199 and @xmath200 ) , the behavior of @xmath98 and @xmath8 rapidly changes at @xmath201 , corresponding to the transition to the normal ( non - superconducting ) state ( @xmath202 ) . in this state ,
@xmath203 does not depend on @xmath23 because setting @xmath202 in ( [ bz_l ] ) and ( [ bz_l1 ] ) gives the free boundary conditions .
the suppression of the superconductivity due to the strong contact with polarization resembles the effect of an external field , impurities and other pair - breaking factors on superconductors@xcite . like these factors , the polarization at the boundaries with the s - layer supplies additional energy ( [ f_fe_s ] ) for the breaking of electron pairs in the strong - coupling regime . from the point of view of the landau - ginzburg functional ( [ f_ferr ] ) , ( [ f_s ] ) , ( [ f_fe_s ] ) , an increase of @xmath91 induces larger values of the fourth - order fluctuation terms as well as the third terms in ( [ f_ferr ] ) and ( [ f_s ] ) describing the spatial inhomogeneities , and hence drives a transition to the homogeneous state with @xmath202 which has the lower energy . to explain qualitatively the effect of @xmath91
, we consider for simplicity only the left fe - s contact with the energy @xmath204 assuming at the right contact the free boundary conditions . at low @xmath205
the expression for the free energy in the ordered fe+s - phase is given by @xmath206+\frac{\gamma}{2}p_b \psi_b^2.\ ] ] the critical value for @xmath91 can be found by equating ( [ f_0 ] ) and the free energy for the state with the suppressed superconductivity @xmath207 which yields @xmath208 from ( [ gamma_c ] ) it follows , that as the s - layer becomes thicker , a stronger critical coupling is required to destroy the superconducting state .
in contrast , a larger polarization @xmath8 in eq .
( [ gamma_c ] ) reduces @xmath209 , which supports our interpretation of the analogy with other external field pair breaking effects .
= 8.5 cm = 8.5 cm consider now the @xmath91-dependence at higher temperatures shown in the insets of fig .
[ fig3](a ) and fig .
[ fig3](b ) slightly below ( @xmath210 ) and above ( @xmath211 , where the system for small @xmath135 is in the domain phase ) the bulk value @xmath212 .
like at low temperatures , the superconductivity of the domain phase ( @xmath211 ) is destroyed when approaching the critical value @xmath213 .
we note that since @xmath202 for @xmath214 , the polarization , which is enhanced here only due to the coupling with @xmath98 , vanishes and we obtain the combined paraelectric+suppressed superconducting ( @xmath153 , @xmath202 ) phase for @xmath215 .
we emphasize that in contrast to the bulk ferroelectrics exhibiting @xmath153 in this temperature range , in our coupled fe - layer the zero polarization is only possible due to the strong coupling effect depressing the source for the enhancement of @xmath8
. the behavior for large @xmath135 can be even more complicated as demonstrated in the inset of fig .
[ fig3 ] at the temperature @xmath210 which is below but close to @xmath216 ( fig .
[ fig3](a ) ) .
despite @xmath217 , the @xmath91-dependence of @xmath8 and @xmath98 clearly shows the stabilization of the domain state with @xmath218 and @xmath219 in the range ( @xmath220 ) , and only with the further @xmath135 increase ( @xmath221 ) , the transition to the ( @xmath222 , @xmath202)-state . to analyze more comprehensively the behavior for different @xmath135
, we discuss in the next section the phase diagrams ( @xmath135 , @xmath198 ) plotted for the different fe- and s - layer thicknesses .
the phase diagrams ( @xmath135 , @xmath198 ) plotted in fig .
[ fig4 ] and fig .
[ fig5 ] show the regions of stability for different phases discussed in section [ phases ] . as has been revealed in the foregoing analysis
, the following phases can exist in our periodical multi - layer system : ( i ) the ordered bulk - type fe+sc phase with the nonzero spontaneous polarization in the fe - layer and the superconductivity in the s - layer ; ( ii ) the superconductivity in the s - layer coexisting with the ferroelectric domain state exhibiting the symmetric ( negative and positive ) polarization domains due to the fe - s contacts ( so called domain - type phase ) ; ( iii ) the phase with the ferroelectricity and the suppressed superconductivity ( fe+nm ) ; ( iv ) the high - temperature paraelectric ( in the fe - layer)+normal state ( in the s - layer ) . in fig .
[ fig4 ] we plot a set of the diagrams for the s - film of moderate thickness @xmath223 coupled to the fe - layers of different thickness : ( a ) a thin fe - film , @xmath224 , ( b ) a fe - layer of moderate thickness @xmath225 , and ( c ) a thick fe - layer with @xmath226 .
we associate in the following analysis the variation of the fe - thickness as related to that of the s - layer , to the ratio @xmath227 where the ratio between the superconductor and the ferroelectric coherence lengths @xmath228 ranges from @xmath229 ( for hts compounds ) to @xmath230 ( conventional superconductors ) , taking the typical value for @xmath231 nm ( ref . ) . as @xmath232 decreases from @xmath233 ( fig .
[ fig4](c ) ) to @xmath234 ( fig .
[ fig4](a ) ) , the area of the domain - type phase drastically expands towards larger @xmath135 and lower @xmath205 and finally approaches even the low - temperature range as represented by arrows in fig .
[ fig4](a ) .
the expansion of the domain phase is accompanied by the simultaneous decrease of the region of the fe+nm(suppressed)-phase .
we note that in the system with a thicker fe - layer ( @xmath235 , fig .
[ fig4](a ) ) , the transition to the suppressed state occurs at a significantly lower @xmath236 ( @xmath237 ) as compared to a thin fe - film ( case @xmath238 in fig .
[ fig4](a ) ) with @xmath239 ( @xmath154 ) , which clearly demonstrates the crucial role of the ferroelectric layer in the suppression of the superconductivity .
these modifications of the phase diagram can be explained by the competition between _ the first tendency _ to destroy the superconductivity prevailing for the coupling with thicker fe - layers , and _ the second tendency _ to retain the superconducting state and stabilize @xmath8-domains dominating in systems with thicker s - layers . as is demonstrated in fig .
[ fig4 ] , the decrease of @xmath232 results in a dominant role of the s - layer and , as a consequence , in the enhancement of the region with @xmath240 at lower @xmath241 ( which in turn induces @xmath8-domains in the fe - layer ) .
since in the hts - films @xmath228 is about three orders of magnitude smaller than in conventional superconductors , one might expect from ( [ zeta ] ) that even a slight variation of the s- or fe - film thickness leads to a dramatic modification of the phase diagram and , consequently , of the stable state of this type of compounds .
= 8.cm fig .
[ fig5 ] shows the behavior of the system with a fixed fe - thickness @xmath235 and with a s - thickness @xmath242 varying from @xmath243 ( fig .
[ fig5](a ) ) to @xmath244 ( fig .
[ fig5](b ) ) . as in the previous case , we find here the same tendency of expansion of the domain phase with decreasing @xmath232 .
we also bring to attention that the destruction of the superconducting state demonstrated here has been observed in the measurements of the resistivity in the film ybco / batio@xmath2 composites @xcite .
= 8.cm as has already been mentioned in section [ model ] , when the ferroelectric critical temperature @xmath22 is close to @xmath4 , the phenomenological approach proposed in this work , has common key aspects with the microscopic description proposed in ref .
however , while the latter model describes the superconducting layer as two boundary 2-d planes exhibiting bcs - superconductivity , the present approach allows us to study the behavior inside the thick s - layers .
despite the fact that the microscopic bcs - model of ref .
does not consider directly the internal processes in the superconducting layer and treats the inner part of the layer as a source for the electron transfer towards ( away from ) the boundary s - planes , the following key results appear in both approaches : + ( i ) the enhancement of the ferroelectric polarization with a formation of positive and negative @xmath8-domains above the ferroelectric transition temperature as well as for large @xmath91 and low temperatures ; + ( ii ) the suppression of the superconductivity for the strong fe - s coupling due to the ferroelectric field - type effect .
+ thus , the basic structure of the phase diagrams ( @xmath91 , @xmath205 ) together with the behavior of the system in different observed phases discussed in ref . , also retains its fundamental features in the present phenomenological approach . the behavior of the different phases discussed above with the variation of the fe- and s - thicknesses results in different types ( from monotonous to complex nonmonotonic ) of the corresponding @xmath193 dependences .
since the system properties drastically change for the large coupling @xmath245 , the @xmath16- and @xmath17-dependence of @xmath193 is also qualitatively different in the weak ( @xmath246 ) and in the strong coupling regime ( @xmath245 ) . to illustrate this , in fig .
[ fig6 ] and fig .
[ fig7 ] we plot several curves for various values of @xmath135 .
specifically , fig .
[ fig6](inset ) and fig .
[ fig7](a ) show the case @xmath246 , whereas @xmath245 is represented by fig .
[ fig6](@xmath247 ) and fig .
[ fig7](b ) . depending on the value of @xmath135
, we can obtain the following characteristic types of @xmath248 and @xmath249 behavior : + ( 1 ) _ for @xmath246 _ , @xmath250 decays monotonically with increasing @xmath16 to the bulk value @xmath212 ( fig .
[ fig6 ] , @xmath251 in the inset ) .
we note that in contrast to the model in ref . where @xmath193 lies slightly below @xmath252 , the coupling with a s - layer of a finite thickness produces here the enhancement of the bulk - type fe+sc - phase .
this corresponds to the increase of @xmath193 ( @xmath253 ) for @xmath106 ( @xmath254 ) , observed also in the phase diagrams in figs .
[ fig4 ] and [ fig5 ] . as a result
, we also obtain that : + ( 2 ) _ for @xmath246 _ , the increase of @xmath17 produces a monotonic increase of @xmath255 , with finally approaching a saturation value which is the higher the larger @xmath135 ( fig .
[ fig7](a ) ) .
+ = 9.cm
\(3 ) _ for @xmath245 _ , the complex behavior of @xmath255 vs @xmath16 shown in fig .
[ fig6](@xmath256 ) , is governed by a decrease of the domain phase - region as @xmath16 increases ( see the plots in fig . [ fig4 ] ) .
first , for thin fe - films , the system is in the domain phase already for @xmath257 . as @xmath258 increases
, the area of the domain state shrinks towards higher @xmath205 and finally disappears , which results in an increase of @xmath255 up to @xmath212 related to the transition from the fe+nm(suppressed ) to the paraelectric state ( see the corresponding diagram in fig .
[ fig4](c ) ) .
note that for a small thickness of the fe - film ( @xmath259 in fig .
[ fig6 ] ) , the behavior of @xmath255 reflects a `` crossover '' between these two different ferroelectric transition types , namely , between the transition from the fe+nm(suppressed ) to the domain state for the thin fe - films , and the transition from the fe+nm to the paraelectric phase for the thick fe - layers .
this effect is also seen in the large @xmath135-region of the diagrams plotted in fig .
+ ( 4 ) fig .
[ fig7](b ) shows a nonmonotonic dependence of @xmath255 on @xmath17 _ for @xmath245_. as in the @xmath16-dependences , for a given value of @xmath135 at a moderate thickness @xmath260 , we find similar , but reversed `` crossover '' between the two above - mentioned different transition types .
specifically , while for thin s - films , the increase of @xmath205 results in the transition from the fe+nm(suppressed ) to the paraelectric state ( at @xmath261 ) , the increase of the s - thickness induces the enhancement of the domain - type phase , leading to a lower @xmath255 ( which now corresponds to the transition between the fe+nm and the domain phases ) . see the evolution of the phase diagram topology as @xmath242 varies , shown in fig .
furthermore , we note that for the larger @xmath262 the transition from the fe+nm- to the symmetric domain - state exhibits : + ( i ) monotonic decay of @xmath255 and vanishing at @xmath263 ( case @xmath264 in fig .
[ fig7](b ) ) ; + ( ii ) reentrant behavior characterized by the vanishing of @xmath255 in a certain interval @xmath242 ( @xmath265 for @xmath266 in fig .
[ fig7](b ) ) and finite @xmath255 otherwise ; + ( iii ) for the larger values of @xmath135 ( @xmath267 in fig .
[ fig7](b ) ) , @xmath255 first decays to a finite value exhibiting a minimum at a particular @xmath17 ( @xmath268 ) , and then increases again with a further increase of the s - thickness , approaching a saturation value for thick s - layers .
this behavior has a simple physical explanation : once the domain state is expanded down to @xmath133 ( yielding @xmath269 at @xmath266 and @xmath270 ) the further increase of @xmath17 for strong coupling leads to the larger spatial inhomogeneous terms in the fe - energy ( [ f_ferr ] ) , which results in a slight enhancement of the suppressed state towards higher @xmath205 and as a result , to an increase of @xmath255 ( see the two phase diagrams for @xmath271 and @xmath272 shown in fig . [ fig5](a ) ) .
in this paper we have developed a phenomenological model describing periodic ferroelectric - superconductor multilayers .
this model generalizes in some aspects the microscopic approach proposed in ref . where the superconducting layers are represented by two planes with a possibility of charge transfer between the planes and the interior of the layer .
we analyze here the case when the ferroelectric transition temperature lies below the temperature of the transition to the superconducting state ( @xmath117 ) which is realized in some complex insulating perovskite alloys such as ba@xmath0sr@xmath1tio@xmath2 ( bst ) coupled to high - temperature superconductors . similarly to the microscopic model , we observe the stabilization of the symmetric domain - type phase exhibiting a weak polarization due to the fe - s contacts above the bulk ferroelectric transition temperature . on the other hand , we find the polarization - induced suppression of the superconductivity in the s - layer .
we show that the ultimate reason for the destruction of the superconducting state for strong fe - s coupling , is the increase of the contribution of the spatial inhomogeneities to the energy of the system .
we study the behavior for different thicknesses of the fe and s - layer demonstrating the strong change of the topology of the phase diagrams ( @xmath135 , @xmath198 ) with a variation of @xmath258 and @xmath242 . in general , we find several types of @xmath248 and @xmath249 behavior .
for the weak fe - s coupling , @xmath246 , we obtain a monotonic variation of @xmath193 , with approaching saturation value in the system with the thick fe or s - layer .
in contrast to this , for the strong coupling , @xmath245 , the dependence of @xmath193 vs layer thicknesses appears to be highly nonmonotonic due to the competition between the first tendency to expand the domain state ( prevailing for the thick s - layers ) and the second tendency to destroy the superconductivity dominating for the thicker fe - layers .
this work has been supported by the schwerpunktprogramm of the german science foundation ( deutsche forschungsgemeinschaft ) under grant no .
let us consider in more details the superconducting layer without a magnetic field .
the general form of the ginzburg - landau functional in the s - layer can be written as @xmath274,\nonumber\end{aligned}\ ] ] where @xmath275 .
note that in the obtained in this case boundary conditions the terms containing @xmath273 and @xmath280 appear with the opposite signs and thus cancel , resulting in the standard form given by ( [ bz_l1 ] ) and ( [ bz_l ] ) in the paper . from ( [ gl_a ] )
we immediately obtain that the single possible solution for @xmath273 is : @xmath281 we assume here for simplicity the absence of the tunneling current through the insulating fe - layer , which corresponds to the zero phase difference on the contacts between the s - films : @xmath282 and leads to the constant phase in the whole system : @xmath283 . as the result , we get from ( [ gl_a ] ) @xmath284 and the next two equations transform into the equation ( [ g_lan_eq2 ] ) for the absolute value of the superconducting order parameter @xmath34 .
vendik , e.k .
hollmann , a.b .
kozyrev , and a.m. prudan , j. superconduct .
* 12 * , 325 ( 1999 ) .
vendik , l.t .
ter - martirosyan , a.i .
dedyk , s.f .
karmanenko , and r.a .
chakalov , ferroelectrics * 144 * , 337 ( 1993 ) .
petrov , e.f .
carlsson , p. larsson , m. friesel , and z.g .
ivanov , j. appl . phys . * 84 * , 3134 ( 1998 ) .
l. ryen , e. olsson , c.n.l .
edvardsson , and u. helmersson , physica c * 304 * , 307 ( 1998 ) .
petrov , z.g .
ivanov , and s.s .
gevorgian , materials sci .
a * 288 * , 231 ( 2000 ) .
van keuls , r.r .
romanofsky , d.y .
bohman , m.d .
winters , f.a .
miranda , c.h .
mueller , r.e .
treece , t.v .
rivkin , and d. galt , appl .
. lett . * 71 * , 3075 ( 1997 ) .
p. konsin and b. sorkin , phys .
b * 58 * , 5795 ( 1998 ) .
p. konsin and t. rd , physica c * 191 * , 469 ( 1992 ) .
jenks and l.r .
testardi , phys .
b * 48 * , 12993 ( 1993 ) .
j. mannhart , supercond .
* 9 * , 49 ( 1996 ) .
larkins , m.y .
avello , j.b .
boyce , and d.k .
fork , ieee trans .
. supercond . *
as-5 * , 3049 ( 1995 ) .
lemanov and a.l .
kholkin , phys . solid state * 36 * , 841 ( 1994 ) .
lemanov , a.l .
kholkin , and a.b .
sharman , jetp lett .
* 56 * , 562 ( 1992 ) .
xi , c. doughty , a. walkenhorst , c. kwon , q. li , and t. venkatesan , phys .
lett . * 68 * , 1240 ( 1992 )
. k. nakajima , k. yokota , h. myoren , j. chen , and t. yamashita , appl .
lett . * 63 * , 684 ( 1993 ) .
j. mannhart , j. strobel , j.g .
bednorz , and ch .
gerber , appl .
* 62 * , 630 ( 1993 ) .
y. watanabe , appl .
* 66 * , 1770 ( 1995 ) .
d. matthey , s. gariglio , c.h .
ahn , and j .-
triscone , physica c * 372 - 376 * , 583 ( 2002 ) .
y. watanabe , y. matsumoto , m. tanamura , h. asami , and a. kato , physica c * 235 - 240 * , 739 ( 1994 ) .
v. muller , c. hucho , and d. maurer , ferroelectrics * 130 * , 45 ( 1992 ) .
a. bussmann - holder , a. simon , and h. buttner , phys .
b * 39 * , 207 ( 1989 ) .
vezzoli , f. craver , m.f .
chen , and w. stanley , ferroelectrics * 130 * , 215 ( 1992 ) .
o.g . vendik and l.t .
ter - martirosyan , phys .
solid state * 36 * , 1778 ( 1994 ) .
n. pavlenko and f. schwabl , phys .
b * 67 * , 094516 ( 2003 ) .
landau and e.m .
lifshitz , _ electrodynamics of continuous media _ , ( pergamon press , oxford , 1960 ) .
artemenko and a.f .
volkov , sov .
* 22 * , 295 ( 1979 ) .
jackson , _ classical electrodynamics _ , ( wiley @xmath285 sons , new york , 1975 ) .
k. kretschmer and k. binder , phys .
b * 20 * , 1065 ( 1979 ) .
m. tinkham , _ introduction to superconductivity _ , ( mcgraw - hill , new york , 1996 ) .
m.e . lines and a.m. glass , _ principles and applications of ferroelectrics and related materials _ , ( clarendon press , oxford , 1977 ) .
a. gajevskis , i. popova , m. kalnberga , e. birks , z. zaula , v. zauls , g. liberts , a. sternberg , c. czurda , e. seidl , and h. weber , ferroelectrics * 131 * , 375 ( 1992 ) . | we present a phenomenological model of periodic ferroelectric - superconductor ( fe - s ) heterostructures containing two alternating ferroelectric and superconducting layers .
the interaction at the fe - s contacts is described as a coupling of the local carrier density of the superconductor with the spontaneous ferroelectric polarization near the fe - s interface .
we obtain a stable symmetric domain - type phase exhibiting a contact - induced polarization and the ferroelectric domain structure at temperatures above the bulk ferroelectric transition temperature . with increasing coupling energy
, we find the appearance of the ferroelectric phase coexisting with the suppressed superconductivity in the s - film .
the system is analyzed for different thicknesses of the fe- and s - films demonstrating the dramatic change of the topology of the phase diagrams with a variation of the layers thickness .
the results are expected to shed light on processes occurring in high - temperature superconducting films grown on perovskite alloy - substrates exhibiting ferroelectric properties at lower temperatures . |
tw hya is a nearby ( @xmath1 pc ) , young ( @xmath2 myr ) star @xcite surrounded by an accretion disk that evinces a large inner hole as judged from the observed spectral energy distribution ( sed ; * ? ? ?
* ) and mid - ir and sub - mm imaging @xcite .
these data can be modeled with an optically thick disk whose inner edge is located @xmath3 au from the central star . while observations of 10 @xmath4 m silicate emission @xcite together with spatially - resolved emission at 2 @xmath4 m @xcite suggest the presence of dust grains with sizes less than a few microns at stellocentric distances @xmath5 au , this inner disk material appears optically thin , and has been estimated to constitute less than a lunar mass @xcite .
such a small amount of material is compatible with the very low near - ir veiling measured previously @xcite .
warm gas has also been detected within the optically thin inner region of tw hya s circumstellar disk @xcite .
in fact , the gas appears to be distributed in a disk with a similar inclination to the optically thick dust disk observed at larger radii ( @xmath67@xmath7 ; * ? ? ? * ; * ? ? ? * ) . the inner disk of tw hya , while optically thin , is clearly not completely devoid of matter . a variety of accretion signatures
have been detected in tw hya , indicating some flow of gaseous material through the inner disk .
observations of h@xmath8 emission , interpreted in the context of a magnetospheric accretion model , indicated @xmath9 m@xmath10 yr@xmath11 @xcite . a shock model fitted to uv and optical photometry indicated a consistent value of @xmath12 m@xmath10 yr@xmath11 @xcite . a similar model fitted to a separate uv spectrophotometric dataset yielded @xmath13 m@xmath10 yr@xmath11 @xcite .
analysis of two additional epochs of uv data separated by approximately two years indicate an accretion rate of @xmath14 m@xmath10 yr@xmath11 on both occasions @xcite .
still higher accretion rates were suggested by modeling of x - ray emission from tw hya : @xmath15 m@xmath10 yr@xmath11 @xcite .
in contrast , a subsequent study of x - ray emission derived an accretion rate of @xmath16 m@xmath10 yr@xmath11 @xcite .
modeling of the c@xmath17 @xmath18 @xmath19 line flux yields @xmath20 m@xmath10 yr@xmath11 @xcite
. while the large dispersion in accretion rate measurements is due in large part to different systematic effects for different techniques , in some cases the same technique used on different epochs ( generally by different investigators ) has yielded different accretion rates . moreover , a study of na d line variability suggested accretion rates varying between @xmath21 and @xmath22 m@xmath10 yr@xmath11 over a 1-yr period @xcite .
thus , it seems that the amount of material in the inner disk may fluctuate significantly over time . in this paper , we present observations of br@xmath0 emission and near - ir veiling .
the two epochs of data show significant variations in br@xmath0 emission strength , but not in veiling .
we use these findings to examine the viability to different mechanisms proposed to maintain the optically thin inner region of tw hya s disk .
we observed tw hya on ut 17 january 2009 and ut 8 may 2009 with the aries instrument at the mmt .
aries receives a corrected beam from the adaptive secondary of the mmt , and performs imaging and spectroscopy from 1 to 5 @xmath4 m wavelengths ( although the 35 @xmath4 m detector is not yet implemented ) .
we used the echelle mode of aries , with a resolving power of @xmath23 .
the slit is @xmath24 , well - matched to the diffraction - limited , ao - corrected resolution of the mmt .
the small slit allows many orders to be placed on the detector simultaneously , and our setup included @xmath25 non - contiguous orders across the @xmath26 and @xmath27 bands .
based on the echelle dispersion , we expect a velocity resolution of approximately 10 km s@xmath11 .
measured fwhms of argon lamp lines show that the actual instrumental velocity resolution of these observations is approximately 14 km s@xmath11 .
tw hya was observed for 10 minutes on 17 january 2009 and for 40 minutes on 8 may 2009 ( both of these are on - source integration times ) . in each epoch , observations of tw hya were interleaved with observations , at similar airmass , of an a1v telluric calibrator star , hip 54682 .
the data were flat - fielded using spectra of an incandescent lamp , and extracted , wavelength calibrated , and telluric corrected , using iraf packages @xcite .
wavelength calibration used telluric absorption lines and the hitran linelist @xcite , and is accurate to 12 km s@xmath11 .
we also obtained imaging data of tw hya and the nearby star twa 7 during both epochs .
twa 7 is a weak - line t tauri star @xcite of similar age to tw hya @xcite .
it exhibits no detectable emission above the stellar photosphere in the near - ir , although it does have long - wavelength excess emission indicative of a debris disk @xcite . due to the low declinations of both tw hya and twa 7 , we did not have time to obtain spectra of twa 7 .
however , since both sources were observed at similar airmasses and have similar colors , twa 7 can be used for relative photometric calibration . in january , 2009 , both objects were observed using the @xmath28 filter with integration times of 0.9 s. in may , 2009 , the atmospheric transparency was better , and these objects saturated the detector when the @xmath28 filter was used , even for the shortest integration times .
we therefore observed tw hya and twa 7 with a narrowband h@xmath29 filter , using 5 s integrations , during the may observations . during our may observations
, we also obtained spectra of gj 568 a , an m3 v star @xcite .
the total integration time was 25 minutes .
the spectra were calibrated using the same procedure described above for tw hya .
we use gj 568 below for a rough comparison of the veiling observed in tw hya with that of an unveiled main - sequence star .
for each observed epoch , we computed the counts for both tw hya and twa 7 within an aperture of 40-pixel ( @xmath30 ) radius .
the sky value was determined as the median of pixels with radii between 60 and 80 ; this value was subtracted from each pixel in the photometric aperture . assuming the 2mass magnitude of twa 7 , @xmath31
, we derive the @xmath27 magnitude of tw hya at both epochs ( table [ tab : results ] ) .
the two measurements are consistent with each other , and with the 2mass magnitude , @xmath32 .
our br@xmath0 spectra are shown in figure [ fig : brg ] .
we measure equivalent widths ( ews ) from these spectra by integrating the line flux between 2.1644 and 2.1668 @xmath4 m ; this range covers all of the br@xmath0 emission seen in both epochs .
the main uncertainty in our ew determination is the continuum level .
we estimate this uncertainty by fitting the continuum in different spectral regions .
we find 12% error bars for our ew measurements .
the lines measured in both epochs are blue - shifted . in the january data , the blue - shift is more pronounced , and a `` blue shoulder '' is more prominent in the spectrum .
blue - shifts and blue - shoulder - profiles are expected for models of infalling gas , since absorption of br@xmath0 emission by the infalling gas along the line of sight preferentially removes red emission ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the emission in both epochs is substantially broader and more asymmetric than expected from stellar flares , which typically show strong ( h@xmath8 line - to - continuum ratios @xmath33 ) , narrow ( fwhm @xmath34 km s@xmath11 ) , symmetric emission line cores ( e.g. , * ? ? ?
* ; * ? ? ?
similarly , quiescent chromospheric activity produces emission that is narrower and more symmetric than emission produced by accretion processes @xcite . since the difference spectrum for the two epochs is broad and asymmetric lacking a strong , narrow core
the change between spectra is probably not due to chromospheric activity . using the broad - band magnitude of tw hya derived in
[ sec : phot ] , we can convert the observed br@xmath0 ews into line luminosities .
these , in turn , can be converted into accretion luminosities using an empirical relationship derived by @xcite . to convert the accretion luminosities into mass accretion rates requires knowledge of the stellar parameters : @xmath35 . for this calculation
, we use the stellar parameters derived by @xcite , assuming a distance of 56 pc : @xmath36 m@xmath10 , @xmath37 r@xmath10 , and @xmath38 k. these are the same values used by @xcite , enabling a fair comparison with their results .
derived line luminosities , accretion luminosities , and accretion rates are listed in table [ tab : results ] .
we estimate continuum veiling in the @xmath27-band using mg and al lines near 2.11 @xmath4 m .
these lines are found in a spectral region relatively free of telluric lines ( unlike , for instance , the co ro - vibrational bandheads ) , and have been used in past studies for veiling measurements of young stars ( e.g. , * ? ? ?
* ; * ? ? ?
the basic method we use to estimate veiling is to compare the observed spectrum to a synthetic spectrum for a stellar photospheric model .
we first determine oscillator strengths for the mg and al transitions by comparing synthetic nextgen spectra for a solar - type star to observed solar spectra .
synthetic spectra relevant for tw hya are then computed using stellar parameters determined from previous high dispersion optical spectroscopy @xcite : @xmath39 k , @xmath40 , [ m / h]@xmath41 , and @xmath42 km s@xmath11
. the previously derived value of @xmath43 is smaller than the velocity resolution of our data .
we rotationally broaden the synthetic spectra by @xmath44 km s@xmath11 , which is the convolution of the true @xmath45 ( @xmath46 km s@xmath11 ) and the instrumental resolution ( @xmath47 km s@xmath11 ) . finally , we add continuum veiling to these rotationally broadened synthetic spectra and find the veiling value that provides the best match between observed and synthetic data .
fitted veilings are @xmath48 and @xmath49 in january and may , respectively .
the two measurements are thus consistent within the 1-@xmath50 statistical errors .
@xcite used this procedure to derive veilings from nirspec data .
they tested the method with observations of unveiled mk standard stars , and found that a correction was needed to produce zero veilings for these stars .
they suggested that scattered light inside the nirspec instrument may have led to an instrumental veiling contribution .
since the aries spectrograph has not been characterized in this way , we need to test for such systematic effects . using our observed spectrum of gj 568 , which is an unveiled m3v main - sequence star @xcite
, we can constrain any possible instrumental veiling .
we estimate the veiling for gj 568 using the procedure described above with @xmath51 k , @xmath52 , and @xmath44 km s@xmath11 ( figure [ fig : gj568veiling ] ) .
we find @xmath53 .
if we assume that the instrumental veiling is independent of spectral type , at least across the m3 to k7 range , then we can use the derived veiling for gj 568 a to correct the veilings inferred for tw hya .
the corrected veilings for tw hya are @xmath54 and @xmath55 in january and may , respectively .
the inferred veilings in the two epochs are consistent with each other and with with a previous measurement by @xcite .
the variation in br@xmath0 emission line - strength between our two observed epochs , and between previous measurements in the literature , suggests that the rate at which gas is being accreted by the central star is changing with time .
however , the amount of dusty material in the inner disk does not appear to change significantly , based on the inferred veilings .
our measured photometry is consistent with the 2mass magnitude , providing further evidence that the amount of matter in the inner disk of tw hya is not changing substantially with time .
several mechanisms for creating the optically thin clearing in the tw hya disk have been proffered ( e.g. , * ? ? ?
* ) , including planets , which may clear gaps about their orbits @xcite ; photoevaporation of inner disk material ( e.g. , * ? ? ?
* ; * ? ? ?
* ) ; or grain growth , which would deplete the population of small grains that would produce the near - ir emission . the accretion rates measured here are difficult to reconcile with euv - driven photoevaporation scenarios @xcite , although not necessarily with euv+x - ray - driven photoevaporation models @xcite .
the sharp edge of the tw hya disk ( e.g. , * ? ? ?
* ; * ? ? ?
* ) seems to argue against grain growth .
while a planetary explanation may be favored , it too is problematic since the massive outer disk of tw hya should cause rapid migration and destruction of planet - mass objects unless the disk is unusually inviscid ( e.g. , * ? ? ?
* ) . while a 10 jupiter - mass planet has been claimed to orbit tw hya at 0.04 au @xcite , this would not dynamically effect the outer disk edge at @xmath3 au
moreover , follow - up observations demonstrate that the claimed planetary signal is more likely due to starspot noise @xcite .
any of the mechanisms discussed above are compatible with time - variable accretion .
time - variable accretion in the outer disk for example , due to changes in mri turbulence can lead to a larger flow of material through the inner disk , even if the flow is impeded by a planet or photoevaporation .
in fact , the inner disk accretion rate can vary indpendently since it too may be driven by mri turbulence @xcite . while our data indicate a changing accretion rate with time , the low veilings observed at both epochs indicate a small and nearly constant amount of dust in the optically thin region . a constant dust content with time
is also compatible with the unchanging @xmath27-band magnitude inferred from our data .
in fact , the lack of optically thick dust in the inner disk at any observed epoch is surprising if one makes a simple estimate of the disk column density .
for this calculation , we assume @xmath56 and @xmath57 .
we take the disk scale height to be a constant fraction of the radius , @xmath58 .
the pressure in the disk is thus @xmath59 . for the accretion rates derived here ( @xmath21 m@xmath10 yr@xmath11 ) , and assuming a viscous disk with @xmath60
, we derive a disk surface density at @xmath61 au of @xmath62 and a disk density of @xmath63 if we assume @xmath64 ( appropriate for sub - micron - sized dust ; see , e.g. , * ? ? ?
* ) , and take a gas - to - dust ratio of 100 , then the vertical optical depth of the disk at 1 au is @xmath65 even for accretion rates lower by an order of magnitude , such as those inferred in previous studied ( section [ sec : intro ] ) , @xmath66 for this simple , viscous disk calculation . to explain the lack of dust in the inner disk , we propose three possible scenarios , all of which may operate in the tw hya system .
the first is that the density could be over - estimated .
an @xmath8disk may not be the correct description for the inner disk of tw hya , or a higher value of @xmath8 may be appropriate . a gas - to - dust ratio @xmath67 could also help to maintain a higher accretion rate with a smaller column of dusty material .
however , observations suggest that the gas - to - dust ratio in tw hya at least as measured in the outer disk may be substantially _ smaller _ than 100 @xcite .
the second possible explanation is dust - filtering , where the pressure gradient at the inner edge of the optically - thick disk leads to super - keplerian velocities , and hence trapping , of dust particles in a certain particle - size range ( e.g. , * ? ? ?
* ) . by preventing dust from reaching the inner disk , filtering can maintain an optically thin region .
moreover , even if the gas flow through the inner disk varies , filtering may trap ( some of ) the additional dust particles in the flow , preventing variations in observed veiling .
finally , radiation pressure from the star can help to drive dust particles out of the inner disk .
alone , this mechanism is unlikely to clear out the inner disk or tw hya , since an optically thick disk can be replenished by accretion faster than material can be blown out ( e.g. , * ? ? ?
. however , if the inner disk is optically thin to begin with , for example due to dust filtering , then radiation pressure acts directly on all dust in the system and can be an efficient means of removal ( e.g. , * ? ? ?
* ; * ? ? ?
radiation pressure and dust filtering act in a complementary way .
dust filtering is most efficient for particles with sizes of tens of microns , since these are not too well coupled to the gas but do nt have the inertia of larger bodies @xcite .
radiation pressure is most effective for smaller particles , with sizes @xmath68 @xmath4 m ( e.g. , * ? ? ? * ) .
thus , small particles that pass through the `` filter '' into the inner disk may be pushed back out by radiation pressure .
the combination of filtering and radiation pressure may be sufficiently efficient to maintain an inner disk virtually devoid of dust even during periods of enhanced gaseous accretion . , j. r. , carr , j. s. , glassgold , a. e. , & valenti , j. a. 2007 , in protostars and planets v , b. reipurth , d. jewitt , and k. keil ( eds . ) , university of arizona press , tucson , 951 pp .
, 2007 . , p.507 - 522 , ed . b. reipurth , d. jewitt , & k. keil , 507522 , l. s. , jacquemart , d. , barbe , a. , benner , d. c. , birk , m. , brown , l. r. , carleer , m. r. , chackerian , c. , chance , k. , coudert , l. h. , dana , v. , devi , v. m. , flaud , j. m. , gamache , r. r. , goldman , a. , hartmann , j. m. , jucks , k. w. , maki , a. g. , mandin , j. y. , massie , s. t. , orphal , j. , perrin , a. , rinsland , c. p. , smith , m. a. h. , tennyson , j. , tolchenov , r. n. , toth , r. a. , vander auwera , j. , varanasi , p. , & wagner , g. 2005 , journal of quantitative spectroscopy and radiative transfer , 96 , 139 , k. i. , calvet , n. , hartmann , l. , kemper , f. , forrest , w. j. , watson , d. m. , dalessio , p. , chen , c. h. , furlan , e. , sargent , b. , brandl , b. r. , herter , t. l. , morris , p. , myers , p. c. , najita , j. , sloan , g. c. , barry , d. j. , green , j. , keller , l. d. , & hall , p. 2004
, , 154 , 439 | we present two epochs of observations of tw hya from the high - dispersion near - ir spectrograph aries at the mmt .
we detect strong emission from the br@xmath0 transition of hydrogen , indicating an accretion rate substantially larger than previously estimated using hydrogen line emission . the br@xmath0 line - strength varies across our two observed epochs .
we also measure circumstellar - to - stellar flux ratios ( i.e. , veilings ) that appear close to zero in both epochs .
these findings suggest that tw hya experiences episodes of enhanced accretion while the inner disk remains largely devoid of dust .
we discuss several physical mechanisms that may explain these observations . |
the discovery of high - temperature superconductivity has motivated a considerable effort in physics of strongly correlated electronic systems , and many theoretical models have been proposed@xcite .
the hubbard model @xcite is one of the simplest and more general description of an interacting electrons system on a lattice . in its simplest form , it includ the usual kinetic energy of electrons and the competition between the on - site electron - electron interaction .
the standard hubbard model is defined by the second quantized hamiltonian @xcite : @xmath0 the first term of the eqs(1 ) represents the kinetic energy of electrons , and each electron has a possibility of hopping between different lattice sites .
@xmath1 is the annihilation operator of the electron at a lattice site @xmath2 with spin index @xmath3 .
@xmath4is the creation operator of the electron at a lattice site @xmath5 , so @xmath6 is the hopping integral from the site @xmath2 to the site @xmath5 .
the second term represents the intrasite coloumb interaction with energy @xmath7 , where @xmath8 is the number operator of electrons at the site @xmath5 with spin @xmath3 .
recently , the random phase approximation ( rpa)@xcite , was used to resolve the standard hubbard model@xcite .
the rpa is an approach which treats seriously the correlations of system , and attempt to minimise the system energy .
the standard hamiltonian hubbard must be developed as function of the creation and annihilation operators of the pair particle - hole(p - h ) , because our rpa regroups the electrons system on pair : particle - hole .
the application of rpa to standard hubbard gives non linear coupled equations , where the resolution is done by a scrpa@xcite . in this paper
, we consider an extended hubbard model@xcite , where the intersite coulomb interactions are introduced .
this model was shown to describe many interesting properties of high tc superconductors materials @xcite . in this work ,
we apply the scrpa method to this extended hubbard model , and study the intersite interaction effect on the ground state and excitation energy of system .
we show that the repulsive interaction between the electrons of the neighbouring atoms is the origin of a supplementary conductivity of the system .
the standard hubbard model with intrasite interaction explains some important physical phenomena like high - temperature superconductivity beenen , mott - transition@xcite . to explain other physical phenomena observed in different areas of the solid state physics like magnetic and transport properties , it is convenient to take into account also the intersite interaction resulting from original coloumb repulsion modified by the polaronic effect .
the extended hubbard hamiltonian is then given bycalegari , grabiec : @xmath9 u denotes the effective intrasite coloumb interaction .
@xmath10 ( @xmath11 ) describes the effective intersite coloumb interaction between the electrons in the lattice sites i and j , with the same spins ( opposite spins ) . @xmath10 and @xmath12 are not necessary equal .
the model(2 ) can not be solved in a general case .
there is however , a special but non trivial case of finite number of sites , which possesses exact analytical solution@xcite . in this work
we will limit ourselves to a simple case , and will apply the general formalism of scrpa to the two sites problem .
we consider a closed chain in one dimension , with periodic boundary conditions @xmath13 .
our physical system is then equivalent to two neighbouring atoms containing a free electron .
the hamiltonian of the system is : @xmath14 where @xmath15 . in order to apply the approximation scrpa to the hubbard model ,
it is necessary , first , to apply the hartree - fock approximation(hf ) to the hubbard model . in the hf method
, we write the hamiltonian(3 ) as function of quasiparticles operators , wich allow us to obtain the excitation spectrum of independent quasiparticles .
the states @xmath16 are defined as : @xmath17 , where @xmath18 is the annihilation operator of the mode @xmath19 ; @xmath20 is related to @xmath21with the usual fourier transformation : @xmath22 k is the momentum of state @xmath23 .
the periodic boundary conditions suppose that @xmath24 . with this condition , eqs(4 )
gives @xmath25 , which have two solutions in the first brillouin zone : @xmath26 and @xmath27 .
thus the hamiltonian is then written as : @xmath28 this expression shows that in the hf approximation , the physical system has two possible states @xmath16 and @xmath29 . @xmath16 ( @xmath29 ) is the hartree - fock ground state ( excited state ) , with the momentum : @xmath30 below ( and @xmath31 above ) the fermi momentum , and the excitation energy : @xmath32 ( and @xmath33 ) , where @xmath34 and @xmath35 . as in ref @xcite , we define the hf quasiparticle operators by : @xmath36 and @xmath37 .
we have then @xmath38 , for all k. in normal ordering of @xmath39 and @xmath40 , the hamiltonian@xmath41 becomes : @xmath42 where @xmath43
@xmath44 with @xmath45 @xmath46 @xmath47 and @xmath48 take into account the correlation between the number operators of the type : @xmath49 in the ground state : @xmath26 ( below the fermi momentum ) and between the magnetic momentum operators of the type : @xmath50 in the excited state : @xmath27 ( above the fermi momentum).@xmath51
in order to apply the formalism of scrpa to the hubbard model , it is convenient to use the particle - hole(ph - rpa ) approximation , which regroup the physical system on pair .
we can then define the rpa excitation operator as : @xmath52 where @xmath53 ( and @xmath54 ) are the momentum below ( and above ) the fermi momentum .
eqs(7 ) shows that the excitation in the ph - rpa is done only by the creation or ( annihilation ) of pair : particle - hole via the operator @xmath55 ( @xmath56 ) with the amplitude @xmath57 ( @xmath58 ) .
the corresponding excited state of this excitation operator is @xmath59 , and the corresponding excitation energy is : @xmath60 \right ] \left| rpa\right\rangle } { \left\langle rpa\right| \left [ q_{v},q_{v}^{^{\dagger } } \right ] \left| rpa\right\rangle } \ ] ] where @xmath61 is the vacuum of this rpa excitation operator : @xmath62 the minimization of @xmath63 leads to usual rpa equations of type : @xmath64 with the relations of the orthonormality conditions of the set @xmath65 , we can express the elements of @xmath66 and @xmath67 by the rpa - amplitudes , and therefore we have a completely closed system of equations for amplitudes @xmath68 and @xmath69 . for our problem , we consider only the excitation operators , which conserve the spin , where the excitation is done only by the creation or annihilation of the pair : particle - hole with the same spin .
@xmath70 with @xmath71 , @xmath72 and @xmath73 , where the mean values @xmath74 are taken with respect to the rpa vacuum @xmath75 .
the scrpa equation can then be written in the form : @xmath76 where the scrpa matrix elements are given by : @xmath77 \right ] \right\rangle\ ] ] and @xmath78 \right ] \right\rangle\ ] ] the orthonormality relations of the set @xmath65 , give : @xmath79 from the hamiltonian given in eqs(10 ) , we can writte the scrpa matrix elements as : @xmath80 , @xmath81 , where : @xmath82 and @xmath83 where @xmath84
so , the ph - rpa matrix takes the form : @xmath85 this ph - rpa matrix has two positive roots : @xmath86 the corresponding eigenvectors are : @xmath87 \ $ ] and@xmath88 $ ] , respectively . where @xmath89 and @xmath90 so , like the hf approximation , in ph - rpa , our system have tow excitation energies @xmath32and @xmath91 , but they are coupled .
thus in this work , we solve a system of a coupled equations numerically by iteration leading to a scrpq solution which are quasi identical to the exact result . to show the effect of intersite interaction on the energy of the system , we have studied th evolution of the ground state and excited energies in term of the interaction @xmath92 and @xmath93 .
in figure1 we plot the variation of the ground state energy @xmath94 as function of the two parameters of the intersite interaction @xmath92 and @xmath95 the result shows that the ground state energy is quasi independent on @xmath92 but , the variation of @xmath96 become more important when we introduce the intersite interaction with the opposite spins : @xmath93 .
this results can be explained by the fact that the scrpa include only , for the fundamental state , the correlations between the particles with different spins * * : * * @xmath97 thus the only type of interaction wich is of interest is @xmath95 in the following we analyse the @xmath93 effect on the dynamics of system . in figure 2 and 3
we plot the variation of the gorund state energy and the excitation energies of scrpa , respectively , as function of the repulsive intrasite interaction u for different values of the intersite interaction @xmath95 the results show that , when @xmath7 increases , the gap @xmath98 between the two excitation energies increases too , and so , the jump of electrons between these two states becomes more difficult .
thus , we can conclude , that for a fixed value of u , while the intersite interaction is repulsive ( attractive ) , the gap @xmath99 becomes less ( more ) important .
these remarks allow us to assume that repulsive interaction between the electrons of the neighbouring atoms is the origin of supplementary conductivity of the system .
in this paper , the scrpa approximation was used to solve the extended hubbard model given in eqs(3 ) .
the quality of the scrpa method has been investigated in a previous work by jemai@xcite , in which he has shown a remarkable agreement between scrpa method and excact results for the standard hubbard model . in our work , we have extended this technic to study the intersite interaction effects on the dynamics of the electrons in the two sites with @xmath100 .
we have shown that the gap between the excitation energies : @xmath101 are correlated with the intersite interaction energy @xmath93 .
this result allow us to suppose that the repulsive intersite interaction ( between the electrons of the neighbouring atoms ) is the origin of a supplementary conductivity of the system .
in future work@xcite , we propose to solve the 4-sites case ( plaquette ) , which may be very important for the explanation of high @xmath102 superconductivity , by considering the many plaquette configurations in 2d .
99 bernhard edegger , noboru fukushima , claudius gros , v.n .
`` particle number renormalization in almost half filled mott hubbard superconductors '' phys .
b 72 , 134504 ( 2005)_,qingshan yuan , feng yuan , c. s. ting _ `` doping dependence of the electron - doped cuprate superconductors from the antiferromagnetic properties of the hubbard model '' _ _ phys . rev .
b 72 , 054504 ( 2005)_,n . stefanakis _
`` local density of states for the corner geometry interface of d - wave superconductors , within the extended hubbard model '' _ _ phys .
b , 66 ( 2002 ) 024514 _ , ivar martin , gerardo ortiz , a. v. balatsky , a. r. bishop _ `` striped superconductors in the extended hubbard model '' _ _ europhys .
lett . , v. 56 , 849 ( 2001 )
_ i. v. solovyev _
`` optimized effective potential for extended hubbard model '' _ _ phys .
b 60 , 8550 ( 1999 ) _ , m. aichhorn , h. g. evertz , w. von der linden , m. potthoff _ _ `` charge ordering in extended hubbard models : variational cluster approach '' _ _ _ prb 70 , 235107 ( 2004 ) _ , n. stefanakis `` local density of states for the corner geometry interface of d - wave superconductors , within the extended hubbard model '' _ phys . rev .
b , 66 ( 2002 ) 024514 _ , ivar martin , gerardo ortiz , a. v. balatsky , a. r. bishop `` striped superconductors in the extended hubbard model '' _ _ europhys . lett . , v. 56 , 849 ( 2001),__zs .
szabo , zs .
`` superconductivity in the extended hubbard model with more than nearest - neighbour contributions '' _
_ phil . mag .
b , vol.76 , no.6 , 911 ( 1997 ) , _
evandro v.l.de mello _ `` a model for high temperature superconductors using the extended hubbard model''_(cond - mat/9506096 ) , e. v. l. de mello and c. acha _ extended hubbard model applied to study the pressure effects in high - temperature superconductors _ | using the scrpa , we study the intersite interaction effect on the dynamics of n electrons system . we have considered an extended hubbard model including intrasite and intersite interactions , and we have applied this model to a system of two neighbouring atoms containing a free electron .
the application of scrpa to this model allows us to study the intersite interaction effect on the ground state and the excitation energies of system .
we show that the repulsive interaction between the electrons of the neighbouring atoms is the origin of an supplementary conductivity of the system .
* keywords : * extended hubbard model , scrpa , ph - rpa * pacs numbers : * 71.10 , -w.75.10.jm , 72.15.nj |
originally , `` symmetric functions '' are thought of as `` functions of the roots of some polynomial '' @xcite . the factorization formula @xmath0 where @xmath1 is the ( multi-)set of roots of @xmath2 ( a polynomial ) invites to consider @xmath3 as a `` multiset ( endo)functor '' rather than a function @xmath4 ( @xmath5 is a field where @xmath2 splits ) .
but , here , @xmath6 whenever @xmath7 and one would like to get the universal formulas _
i.e. _ which hold true whatever the cardinality of @xmath8 .
this set of formulas is obtained as soon as the alphabet is infinite and , there , this calculus appears as an art of computing symmetric functions without using any variable . with this point of view
, one sees that the algebra of symmetric functions comes equipped with many additional structures @xcite ( comultiplications , @xmath9-ring , transformations of alphabets , internal product , ) . for our concern here ,
the most important of these features is the fact that the ( commutative ) hopf algebra of symmetric functions is self - dual . at the cost of losing self - duality ,
features of the ( hopf ) algebra of symmetric functions carry over to the noncommutative level @xcite .
this loss of self - duality has however a merit : allowing to separate the two sides in the factorization of the diagonal series , thus giving a meaning to what could be considered a complete system of local coordinates for the hausdorff group of the stuffle hopf algebra .
indeed , the elements of the hausdorff group of the ( shuffle or stuffle ) algebras excatly are , through the isomorphism @xmath10 , the characters of the algebra .
then , applying @xmath11 ( @xmath12 ) to the factorization @xmath13 and using the fact that @xmath14 is a ( continuous ) character , one gets a decomposition of @xmath14 through this complete system of local coordinates : @xmath15 this fact is better understood when one considers sweedler s dual of the ( shuffle of stuffle ) hopf algebra @xmath16 , which contains as well @xmath17 and its lie algebra , the space of infinitesimal characters .
such a character is here a series @xmath18 such that ( as a linear form ) @xmath19 and one sees from this definition that such a series , as well as the characters , satisfies an identity of the type @xmath20 for some finite double family @xmath21 .
then in ( [ series ] ) , the character @xmath14 is factorized as an ( infinite ) product of _ elementary _ characters .
this shows firstly , one can reconstruct a character from its projection onto the free lie algebra is the projection onto the free lie algebra parallel to the space generated by the non - primitive elements of the pbw basis ] and secondly , we have at hand a resolution of unity from the process @xmath22 and the key point of this resolution is exactly the system of coordinate forms provided by the dual family of any pbw homogeneous basis .
this paper is devoted to a detailed exposition of the machinery and morphisms surrounding this resolution ( equation ( [ resolution ] ) ) and it is structured as follows : in section [ background ] , we give a reminder on noncommutative symmetric and quasi - symmetric functions , in section [ factorization ] , we focus on the combinatorial aspects of the quasi - shuffle hopf algebra will be introduced to obtain , via schtzenberger s monoidal factorization , a pair of bases in duality for the noncommutative symmetric and quasi - symmetric functions , encoded by words .
for any _ composition _ @xmath23 of strictly positive integers is an element of the monoid @xmath24 and the empty composition will be denoted here by @xmath25 . ] , called the _ parts _ of @xmath26 , the mirror image of @xmath26 , denoted by @xmath27 , is the composition @xmath28 .
let @xmath29 , the _ length _ and the _ weight _ of @xmath26 are defined respectively as the numbers @xmath30 and @xmath31 . the last part and the product of the partial sum of the entries of @xmath26
are defined respectively as the numbers @xmath32 and @xmath33 .
one defines also @xmath34 let @xmath35 be a composition which is finer than @xmath26 and let @xmath36 be the decomposition of @xmath35 such that , for any @xmath37 .
one defines @xmath38 let @xmath39 be a commutative @xmath40-algebra .
the algebra of noncommutative symmetric functions , denoted by @xmath41 , introduced in @xcite , is the free associative algebra generated by an infinite sequence @xmath42 of non commuting inderminates also called _ complete _ homogenous symmetric functions .
let @xmath43 be another variable commuting with all the @xmath42 . introducing the ordinary generating series @xmath44 other noncommutative symmetric functions can be derived by the following relations @xmath45 where @xmath46 are respectively the following ordinary generating series @xmath47 the noncommutative symmetric functions @xmath48 are called _ elementary _ functions .
the elements @xmath49 ( resp .
@xmath50 ) are called _ power sums of the first kind _ ( resp .
_ second kind _ ) .
let @xmath51 , one defines the products of complete and elementary symmetric functions , and the products of power sums as follows @xcite @xmath52 and it is established that @xmath53 the families @xmath54 , @xmath55 and @xmath56 are then homogeneous bases of @xmath57 .
recall that @xmath58 .
the @xmath59-algebra @xmath57 possesses a finite - dimensional grading by the weight function defined , for any composition @xmath60 , by the number @xmath61 .
its homogeneous component of weight @xmath62 ( free and finite - dimensional ) will be denoted by @xmath63 and one has @xmath64 one can also endow @xmath65 with a structure of hopf algebra , the coproduct @xmath66 being defined by one of the following equivalent formulae , with the convention that @xmath67 and @xmath68 @xcite @xmath69 in other words , for the coproduct @xmath66 , the power sums of the first kind @xmath49 and of the second kind @xmath50 are primitive .
the noncommutative symmetric function @xmath70 is primitive but @xmath71 and @xmath72 are neither primitive nor group - like .
moreover , by ( [ 10 ] ) , ( [ 12 ] ) and ( [ 13 ] ) , one has @xmath73 with the concatenation
, the coproduct @xmath66 and the counit @xmath74 defined by @xmath75 one gets the bialgebra , @xmath76 , over the @xmath39-algebra @xmath57 .
this algebra , @xmath77-graded by the weight is , as we will see in theorem [ isomorphisms ] , the _ concatenation _ hopf algebra .
let us consider also an infinite sequence @xmath78 of non commuting inderminates generating the free associative algebra to warn the reader that the structure of free algebra is used to construct the basis of @xmath79 which will be later free _ as a commutative algebra _
( with the stuffle product ) and by no means as a noncommutative algebra ( with the concatenation product ) . ]
@xmath80 and define the elements @xmath81 as follows @xmath82 the elements @xmath81 of @xmath79 are also called the _ monomial quasi - symmetric _ functions .
they are homogeneous polynomials of degree @xmath83 .
this family is then an homogeneous basis of @xmath79 . with the pairing @xmath84 one constructs the bialgebra dual to @xmath57 , @xmath85 , over the @xmath39-algebra @xmath79 . here
the coproduct @xmath86 is defined by @xmath87 2 .
the counit @xmath88 is defined by @xmath89 3 . the product @xmath90 is the commutative product associated to the coproduct @xmath66 and is defined , for any composition @xmath91 , by @xmath92 and for any composition @xmath93 and @xmath94 @xmath95 since the bialgebra @xmath79 is @xmath77-graded by the weight ( as the dual of the @xmath77-graded bialgebra @xmath57 ) : @xmath96 then it is , in fact , the _ convolution _ hopf algebra .
indeed , one can check that , for any @xmath97 , @xmath98
let @xmath99 be a totally ordered alphabet . ] . the free monoid and the set of lyndon words , over @xmath100 , are denoted respectively by @xmath101 and @xmath102 @xcite .
the neutral element of @xmath101 is denoted by @xmath103 .
let @xmath104 , the _ length _ and the _ weight _ of @xmath105 are defined respectively as the numbers @xmath106 and @xmath107 .
let us define the commutative product over @xmath108 , denoted by @xmath109 , as follows @xcite @xmath110 or by its associated coproduct , @xmath111 , defined by @xmath112 satisfying , @xmath113 let @xmath114 be equipped by 1 .
the concatenation ( or by its associated coproduct , @xmath86 ) .
the _ shuffle _ product , _
i.e. _ the commutative product defined by @xcite @xmath115 or by its associated coproduct , @xmath116 , defined , on the letters , by @xmath117 and extended by morphism .
it satisfies @xmath118 3 . the _ quasi - shuffle _ product , _
i.e. _ the commutative product defined by @xcite , for any @xmath119 , @xmath120 and , for any @xmath121 , @xmath122 or by its associated coproduct , @xmath123 , defined , on the letters , by @xmath124 and extended by morphism .
it satisfies @xmath125 note that @xmath123 and @xmath116 are morphisms for the concatenation ( by definition ) whereas @xmath111 is not a morphism for the product of @xmath108 ( for example @xmath126 , whereas @xmath127 ) .
hence , with the counit @xmath128 defined by @xmath129 one gets two pairs of mutually dual bialgebras @xmath130 let us then consider the following diagonal series @xmath131 here , in @xmath132 and @xmath133 , the operation on the right factor of the tensor product is the concatenation , and the operation on the left factor is the shuffle and the quasi - shuffle , respectively . by the cartier - quillen - milnor - moore theorem ( see @xcite ) , the connected @xmath77-graded , co - commutative hopf algebra @xmath134 is isomorphic to the enveloping algebra of the lie algebra of its primitive elements which is equal to @xmath135 : @xmath136 hence , let us consider 1 .
the pbw - lyndon basis @xmath137 for @xmath138 constructed recursively as follows @xcite @xmath139&\mbox{for } l\in\lyn y,\mbox { standard factorization of } l=(s , r),\\ p_{w}&=&p_{l_1}^{i_1}\ldots p_{l_k}^{i_k}&\mbox{for } w = l_1^{i_1}\ldots l_k^{i_k},l_1>\ldots > l_k , l_1\ldots , l_k\in\lyn y , \end{array}\right.\ ] ] 2 . and , by duality , the linear basis @xmath140 for @xmath141 , _
i.e. _ @xmath142 it can be shown that this basis can be computed recursively as follows @xcite @xmath143 hence , we get schtzenberger s factorization of @xmath132 @xmath144 similarly , by the cartier - quillen - milnor - moore theorem ( see @xcite ) , the connected @xmath77-graded , co - commutative hopf algebra @xmath145 is isomorphic to the enveloping algebra of its primitive elements : @xmath146 where , for any @xmath147 is obtained as follows @xcite @xmath148 note that the eq .
[ pi1 ] is equivalent to the following identity which will be used later on @xmath149 in particular , for any @xmath150 , the primitive polynomial @xmath151 is given by @xmath152 as previously , ( [ pi1bis ] ) is equivalent to @xmath153 hence , by introducing the new alphabet @xmath154 , one has @xmath155 by considering 1 .
the pbw - lyndon basis @xmath156 for @xmath157 constructed recursively as follows @xcite @xmath158&\mbox{for } l\in\lyn y,\mbox { standard factorization of } l=(s ,
r),\\ \pi_{w}&=&\pi_{l_1}^{i_1}\ldots\pi_{l_k}^{i_k}&\mbox{for } w = l_1^{i_1}\ldots l_k^{i_k},l_1>\ldots > l_k , l_1\ldots , l_k\in\lyn y , \end{array}\right.\ ] ] 2 . and , by duality , the linear basis @xmath159 for @xmath160 , _
i.e. _ @xmath161 it can be shown that this basis can be computed recursively as follows @xcite @xmath162 we get the following extended schtzenberger s factorization of @xmath133 @xcite @xmath163 [ y ] let @xmath164 be the following ordinary generating series of @xmath165 : @xmath166\ ! ] .
\end{array}\ ] ] then @xmath164 is group - like , for the coproduct @xmath123 .
we have successively ( ( here , in order to make complete the correspondence @xmath167 , we put @xmath168 ) @xmath169 it follows then @xmath170 meaning ( with @xmath171 ) that @xmath164 is group - like .
[ logy ] let @xmath172 be the lie algebra generated by @xmath173 .
then we have @xmath174 .
the power series @xmath175\!]$ ] is primitive then by expanding @xmath176 , we get successively @xmath177 by ( [ pi1bis ] ) , we get , for any @xmath178 , @xmath179 and since @xmath180 generates freely @xmath181 @xcite , the expected result follows . in virtue of ( [ pi1bis ] ) and ( [ pi1ter ] )
, we also have @xmath182 [ yinverse ] let us write the ( group - like ) power series @xmath183 and its differentiation as follows @xmath184\ ! ] & \mbox{and } & \dot{\cal y}^{-1}&=&-{\cal y}^{-1}\dot{\cal y}{\cal y}^{-1}. \end{array}\ ] ] then , for any @xmath178 , one has @xmath185 using the identities @xmath186 , the results follow immediately by identification of the coefficients of @xmath187 and by differentiation , respectively .
[ lr ] there exists two ( unique and primitive ) generating series @xmath188 and @xmath189\!]$ ] satisfying @xmath190 . moreover , if @xmath191 then , for any @xmath178 , @xmath192 on the one hand , by proposition [ y ] , one has @xmath193 on the other hand , such generating series exist since @xmath194 hence , identifying the coefficients of @xmath187 in these identities , the expected results follows .
moreover , since @xmath123 commutes with @xmath195 and it is a morphism for the concatenation then @xmath196 hence , @xmath197 and @xmath198 meaning that @xmath188 and @xmath199 are primitive .
more generally , with the notations of corollary [ lr ] , one has for any @xmath200 , there exist two unique generating series @xmath201\!]$ ] such that @xmath202 .
the families @xmath203 and @xmath204 are defined recursively as follows @xmath205 on the one hand , by proposition [ y ] , one has , as in corollary [ lr ] , @xmath206 where @xmath207 is the pochhammer symbol . on the other hand , by induction * for @xmath208
, it is corollary [ lr ] .
* suppose the property holds for any @xmath209 .
* for @xmath210 , such generating series exist since , by induction hypothesis , @xmath211 hence , @xmath212 and @xmath213 . for any proper power series @xmath214 , let @xmath215 be the iterated lie brackets defined recursively by @xmath216 and @xmath217 $ ] , for @xmath178
. then , with notations of corollary [ lr ] , one has @xmath218 since @xmath219 then @xmath220 and then @xmath221 .
expanding @xmath222 , the results follow .
[ prim ] let @xmath172 be the lie algebra generated by @xmath223 ( resp .
@xmath224 ) . then @xmath174 . by corollary [ lr ] , one has on the one hand , @xmath225 thus , by identifying the coefficients of @xmath226 in the first and last sums , on has @xmath227 , meaning that @xmath228 is primitive . on the other hand , according to basic properties of quasi - determinants ( @xcite , see also @xcite ) , one has @xmath229 hence , for any @xmath230 , by denoting @xmath231 , one obtains @xmath232 it means @xmath233 is triangular and homogeneous in weight in @xmath234 .
conservely , @xmath228 is also triangular and homogeneous in weight in @xmath235 .
the @xmath236 s are then linearly independent and constitute a new alphabet . in the same way , the @xmath237 s are primitive and linearly independent .
the expected results follow .
let us define the families @xmath238 , for @xmath239 or @xmath199 , of @xmath145 as follows @xmath240,&\mbox{for } l\in\lyn y,\mbox { with standard factorization of } l=(s , r),\\ \pi^{(s)}_{w}&=&(\pi^{(s)}_{l_1})^{i_1}\ldots(\pi^{(s)}_{l_k})^{i_k } , & \mbox{for } w = l_1^{i_1}\ldots
l_k^{i_k},l_1>\ldots > l_k , l_1\ldots , l_k\in\lyn y. \end{array}\right.\ ] ] then , the families @xmath241 ( resp .
@xmath238 ) , for @xmath239 or @xmath199 , are bases of @xmath181 ( resp .
@xmath145 ) , these bases are homogeneous in weight . these results ( homogeneity , primitivity , linear independence ) can be proved by induction on the length of lyndon words .
let @xmath242 be the family of @xmath243 obtained by duality with @xmath238 : @xmath244 1 .
the family @xmath245 forms a basis of the lie algebra generated by @xmath181 .
2 . the family @xmath238 forms a basis of @xmath246 .
the family @xmath242 freely generates the quasi - shuffle algebra .
4 . the family @xmath247 forms a transcendence basis of the quasi - shuffle algebra .
the family @xmath245 of primitive upper triangular homogeneous in weight polynomials is free and the first result follows .
the second is a direct consequence of the poincar - birkhoff - witt theorem . by the cartier - quillen - milnor - moore theorem
, we get the third one and the last one is obtained as a consequence of the constructions of @xmath247 and @xmath242 . @xmath248 note
that any word @xmath104 corresponds one by one to a composition of integers @xmath29 ( and the empty word @xmath103 corresponds to the empty composition @xmath25 ) .
note also that noncommutative symmetric functions and quasi - symmetric functions can be indexed by words in @xmath101 instead of by composions in @xmath24 .
indeed , let @xmath35 be a composition , finer than @xmath26 , associated to the word @xmath249 and let @xmath36 be the decomposition of @xmath35 such that , for any @xmath37 and @xmath250 is associated to the word @xmath251 whose @xmath252 . then @xmath253 is a unique factorization and this will be denoted as a bracketing of the word @xmath249 .
one has * @xmath254 .
* @xmath255 .
* @xmath256 . hence , we can state the following let @xmath257 and @xmath258 be the following linear maps @xmath259 [ isomorphisms ] the maps @xmath257 and @xmath258 are isomorphisms of hopf algebras .
let @xmath260 be the lie algebra generated by @xmath261 .
then , we have @xmath262 the families @xmath263 and @xmath264 are transcendence bases of the free commutative @xmath39-algebra @xmath79 .
let @xmath265 associated to @xmath29 .
then , we have @xmath266 on the one hand , the power series @xmath267 and @xmath268\!]$ ] are summable . on the other hand , by ( [ sigma(t ) ] ) and ( [ psi(t ) ] ) , since @xmath257 is continuous and commutes with @xmath269 , one can deduce @xmath270 thus , the expected result follows immediately . with the notations of ( [ factorization2 ] ) , let us consider the following noncommutative generating series @xmath271 and @xmath272 @xmath273 [ m ] for the coproduct @xmath123 , using ( [ factorization2 ] ) , we obtain 1 . the noncommutative generating series @xmath274 is group - like
the noncommutative generating series @xmath275 is primitive . 1
. it follows friedrichs criterion @xcite .
2 . by using the previous result and by applying the @xmath269 map on the power series @xmath274
, we get the expected result .
@xmath276 the first identity is equivalent to the image of the diagonal series @xmath133 by the tensor @xmath277 .
the second one is then equivalent to the image of @xmath275 by the tensor @xmath278 .
it is also equivalent to the image of @xmath133 by the tensor @xmath279 . finally , using ( [ factorization2 ] )
we deduce the following property which completes the formulae ( 120 ) given in @xcite : we have @xmath280 by theorem [ isomorphisms ] , it is equivalent to the image of the diagonal series @xmath133 by the tensor @xmath281 , or equivalently it is equivalent to the image of the power series @xmath274 by the tensor @xmath278 . note that these formulas are universal for any pair of bases in duality , compatible with monoidal factorization of @xmath101 , and they do not depend on the specific alphabets , usually denoted by @xmath282 and @xmath283 , used to define @xmath284 and @xmath285 .
let @xmath282 be a noncommutative alphabet and @xmath283 a totally ordered commutative alphabet .
the symmetric functions of the noncommutative alphabet @xmath286 are defined by means of@xmath287 let @xmath288 and @xmath289 be two linear bases of @xmath290 and @xmath291 respectively .
the duality of these bases means that @xmath292 typically , the linear basis @xmath288 is the basis of _ ribbon _ schur functions @xmath293 , and , by duality , @xmath289 is the basis of _ quasi - ribbon _ schur functions @xmath294 : @xmath295 = { \displaystyle{\sum_{_}^{j\in({{\mathbb n}}_+)^*}}}\biggl[{\displaystyle{\sum_{_}^{i , j\in({{\mathbb n}}_+)^*\atop i\succeq j}}}m_i(x)\biggr]\mathrm{r}_j(a ) = { \displaystyle{\sum_{_}^{j\in({{\mathbb n}}_+)^*}}}\mathrm{f}_j(x)\mathrm{r}_j(a).\end{aligned}\ ] ] also , if one specializes the alphabets of the quasi - symmetric functions @xmath81 and @xmath296 to the commutative alphabet @xmath297 , then the generating series @xmath298 can be viewed as the image of the diagonal series @xmath133 by the tensor @xmath299 , where @xmath300 , and one has let @xmath301 denotes the totally ordered commutative alphabet @xmath302 .
the complete symmetric functions of the noncommutative alphabet @xmath303 are given by the following ordinary generating series @xmath304 hence , @xmath305s^i(a ) = \sum_{i\in({{\mathbb n}}_+)^*}m_i(x)s^i(a),\end{aligned}\ ] ] by specializing each letter @xmath306 to @xmath307 in the quasi - symmetric function @xmath308 .
once again , the schtzenberger s monoidal factorization plays a central role in the construction of pairs of bases in duality , as exemplified for the ( mutually dual ) hopf algebras of quasi - symmetric functions ( @xmath79 ) and of noncommutative symmetric functions ( @xmath57 ) , obtained as isomorphical images of the quasi - shuffle hopf algebra ( @xmath145 ) and its dual ( @xmath243 ) , by @xmath258 and @xmath257 respectively . .
_ theory of codes _ , academic press ( 1985 ) . .
_ schtzenberger s factorization on the ( completed ) hopf algebra of @xmath309stuffle product_. . _ combinatorics of @xmath310-deformed stuffle hopf algebras_. . _ free differential calculus , iv .
the quotient groups of the lower central series _ , ann . of math . ,
68 ( 1958 ) pp . 8195 .
. _ noncommutative symmetric functions _ , advances in mathematics 112 ( 1995 ) , 218 - 348 .
. _ determinants of matrices over noncommutative rings _ , funct .
appl . , 25 , ( 1991 ) , 91 - 102 .
. _ a theory of noncommutative determinants and characteristic functions of graphs _ , funct
appl . , 26 , ( 1992 )
, 1 - 20 ; publ .
lacim , uqam , montreal , 14 , 1 - 26 . .
_ noncommutative symmetric functions iii : deformations of cauchy and convolution algebras _ , discrete mathematics and theoretical computer science 1 ( 1997 ) , 159216 . . _ noncommutative symmetric functions iv : quantum linear groups and hecke algebras at @xmath311 _ , journal of algebraic combinatorics 6 ( 1997 ) , 339376 . . _ noncommutative symmetric functions v : a degenerate version of @xmath312 _ , internat .
( 1999 ) , 405 - 430 . .
_ noncommutative symmetric functions ii : transformations of alphabets _ , international journal of algebra and computation 7 ( 1997 ) , 181 - 264 . .
_ quasi - shuffle products _ , j. of alg .
cominatorics , 11 ( 2000 ) , pages 49 - 68 . .
_ on a conjecture by pierre cartier about a group of associators _ , to appear ( 2013 ) .
. _ combinatorics on words _ , addison - wesley publishing co. , reading , mass . ,
encyclopedia of mathematics and its applications 17 ( 1983 ) .
. _ free lie algebras _ , london math . soc .
monographs , new series-7 , oxford science publications , ( 1993 ) . .
_ algbres de lie libres et monodes libres _ , lecture notes in mathematics , vol .
691 , springer , 1978 . | in this work , an effective construction , via schtzenberger s monoidal factorization , of dual bases for the non commutative symmetric and quasi - symmetric functions is proposed . * keywords * : non commutative symmetric functions , quasi - symmetric functions , lyndon words , lie elements , monoidal factorization , transcendence bases . |
the efficiency of thermoelectric ( te ) material at temperature _ t _ is evaluated by the figure of merit @xmath7 , where _ s _ is the thermopower , @xmath8 is the electrical resistivity and @xmath9 is the thermal conductivity .
there are two distinct approaches to increasing the _ zt _ : either by @xmath9 reduction or by power factor @xmath10 ) enhancement . techniques such as alloying , `` phonon glass electron crystal '' ( pgec ) approach , and nanostructure engineering have been used to reduce the phonon mean free path , reduce lattice @xmath9 and produce high _ z _ , at or above the room temperature.@xcite reduction of the phonon mean free path has limitations since it can not be reduced below the interatomic spacing and therefore other mechanisms for high _ zt _ are sought after.@xcite tuning the carrier density by doping is often inadequate since lower @xmath8 comes with higher carrier concentration that favors lower _
s_.@xcite the most favorable te electronic structure is the one that has a resonance in the density of states centered about 2 - 3 @xmath11 away from the fermi energy ( @xmath12@xmath13).@xcite kondo insulators ( ki ) represent a close approximation of such an ideal case . in a strongly correlated electron system , and , in particular , in a ki , localized _
f _ or _ d _ states hybridize with conduction electron states leading to the formation of a small hybridization gap .
the density of states just below and just above the hybridization gap becomes very large .
the thermopower is very sensitive to variations in density of states in the vicinity of the @xmath12@xmath13 , hence very large absolute values of @xmath14 @xmath15 100 @xmath4v / k can be expected and are reported in ki.@xcite the materials include not only rare - earth based compounds but also fesi.@xcite fesb@xmath0 crystallizes in _ pnnm _ orthorhombic structure and has been characterized as an example of strongly correlated non - cubic kondo insulator - like material with _
ions.@xcite similar to fesi , heavy fermion states were discovered in fesb@xmath0 by doping - induced metallization.@xcite colossal values of thermopower up to @xmath3 45 mv / k at 10 k and a record high thermoelectric power factor ( tpf ) of @xmath3 2300 @xmath4w / k@xmath1 cm were observed.@xcite this is two orders of magnitude larger than in best bi@xmath0te@xmath16-based materials and one order of magnitude larger than any reported value in correlated electron systems .
crystals in refs .
19 and 21 exhibit semiconductor behavior and small resistivity anisotropy with relatively high resistivity for current directed along all three principal crystalline axes . in contrast
, single crystals used in kondo insulator studies showed high anisotropy of the resistivity : electric transport along the _ a _ and _ b _ axes is semiconducting , while the _ c _ axis is metallic at high temperatures and exhibits metal - insulator transition ( mit ) at 40 k.@xcite here we investigated the tpf in two fesb@xmath0 crystals , with ( crystal 1 ) and without ( crystal 2 ) mit .
we show that large _
s _ enhancement can be attributed to electronic correlations in multiple charge and heat carrying bands .
we also provide evidence for structural origin of mit in correlated electron bands that reduces @xmath8 by several orders of magnitude
around 30 k in crystal 1 and results in a new record high tpf . in crystal 1 above 80
k is consistent with semiconducting thermally - activated transport with an activation gap of 120 k. ( b),(c ) temperature regime(s ) where @xmath17 is dominated by thermally activated transport .
the arrow in ( b ) marks the mit in crystal 1 , where @xmath18/dt$ ] changes sign .
broken lines in both panels are linear fits to activation - type behavior . ]
single crystals of fesb@xmath0 with mit have been prepared as in refs .
16 and 17 by decanting at 650@xmath19c .
single crystals of fesb@xmath0 without mit have been prepared by the method described in refs .
19 and 21 , i.e. , by decanting at 690@xmath19c , after a cool down from high temperatures to 640@xmath19c .
crystals were oriented using laue camera and cut into [email protected]@xmath204.5 ( a@xmath20b@xmath20c ) mm@xmath21 samples for two probe thermopower measurements with two ends soldered on disk - shape leads using indium .
resistivity @xmath17 was determined by a standard four - point ac method .
heat and current transport along the orthorombic _
c _ axis were measured using the quantum design ppms platform .
a magnetic field was applied along the crystallographic _ a _ axis .
sample dimensions were measured with an optical microscope nikon smz-800 with 10 @xmath4 m resolution .
consequently the relative errors on the electrical resistivity and thermal conductivity are 4@xmath22 and for the hall coefficient 2@xmath22 .
since the seebeck coefficient does not depend on the sample geometry the main source of error is the sample uniformity and the accuracy of crystal orientation which introduces the measurement error of up to 5@xmath22 .
the atomic pair distribution ( pdf ) method , based on the total scattering approach , yields structural information on different length scales.@xcite x - ray scattering experiments for pdf analysis were carried out at the 11-id - c beamline of the advanced photon source using high energy beam ( _ e _ = 114.82 kev , @xmath23 = 0.108 @xmath24 , 0.5 mm @xmath20 0.5 mm size ) .
both pdf , _
g(r ) _ , and its algebraic relative , radial distribution function ( rdf ) , _
r(r ) _ , were considered in this study .
experimental pdf , _
g(r ) _ is obtained from the measured reduced total scattering structure factor , _
f(q)=q[s(q)-1 ] _ , via sine fourier transform @xmath25 . in practice , the upper limit of integration is some finite value @xmath26 .
r(r ) _ is obtained from _
g(r ) _ through : @xmath27@xmath28@xmath29@xmath30@xmath31@xmath32 where @xmath32 is the average number density . experimental setup for total scattering x - ray experiments utilized a cryo industries of america cryostat and perkin - elmer image plate detector .
finely ground samples in cylindrical polyimide capillaries were placed in a low - temperature sample changer , and the data for the two samples were successively collected for 4 min . at each temperature in probed range between 5 and 300 k. pdfs were obtained up to @xmath26 = 26 @xmath33 momentum transfer using standard protocols , and intermediate length scale structure modeled over ( 1.7 - 45.0 ) @xmath33 range with _ pnnm _ structural model using the program pdfgui.@xcite preliminary reference neutron total scattering based pdf s were obtained using a time - of - flight hipd instrument at los alamos neutron scatttering center .
the @xmath34 for crystal 1 is metallic down to the onset of mit ( @xmath35 ) , as opposed to crystal 2 [ fig .
1(a ) ] . @xcite above 80 k and up to 300 k , @xmath36 of crystal 2 is insensitive to magnetic field and is semiconducting .
interestingly , the logarithmic derivative , @xmath37/dt$ ] reveals an underdeveloped anomaly at @xmath35 in crystal 2 which points to the intrinsic nature of this temperature scale [ fig .
1(b , c ) ] . arrhenius analysis assuming thermally activated behavior leads to a number of distinct energy scales valid for limited subranges of this temperature interval ( table i ) , allowing only approximate estimation of the gap values .
the absolute value @xmath38/dt|$ ] in the metallic and insulating states is very similar , which is reminiscent of the universality observed near the mit in low - dimensional electron gas systems.@xcite below 5 k electrical transport is dominated by extrinsic impurity states and is governed by a very small activation gap , similar for both crystals .
distinct energy scales and indirect gaps are in agreement with optical spectroscopy studies and _ ab initio _ calculations .
@xcite the large activation gap observed above 80 k in crystal 2 is not observed in the metallic phase of crystal 1 , but its size argues against contribution of impurity states .
.approximate ( see text ) energy gap from electronic transport , the ratio of quasiparticle weights and scattering times in the valence and conduction band @xmath39/[(z{_c}^{2}\tau_{c}^{5/2})/\gamma_{c}$ ] ] ) and @xmath12@xmath13@xmath3ln(@xmath4@xmath40/@xmath4@xmath41 ) for a crystal with mit ( top ) and a crystal with no mit ( bottom row).the huge difference of fitting parameters between crystals 1 and 2 may be rationalized by the fact that the mobility difference of two bands for crystals with mit is 10@xmath42 - 10@xmath43 while the one for crystals without mit is @xmath310 , whereas both show similar large seebeck coefficients . [ cols="^,^,^,^,^ " , ] @xmath44 ( a ) and hall constant ( b ) for crystal 2 . hall mobility ( c ) where the hole ( electron ) carriers are denoted by + ( - ) and carrier concentrations of high and low mobility carriers ( d ) as a function of temperature for both crystals . ] the hall resistivity @xmath45 ( current flowing along the _ c _ axis ) for both crystals [ fig 2 ( a , b ) ] is nonlinear , confirming the presence of two carrier bands . in the absence of skew scattering
we proceed to the analysis of the data in the two - band picture . in a two - carrier system
@xcite the hall coefficient is @xmath46 where @xmath47 , @xmath48 and @xmath49 where @xmath32 = @xmath50 , @xmath51 is the f factor , @xmath52 and @xmath53 are individual carrier band concentrations and mobilities .
the agreement with the model is excellent and we obtain carrier band concentrations and mobilities for crystal 2 while also showing data for crystal 1 for comparison.@xcite in the model `` carrier '' denotes a set of carriers with identical mobility associated with only one energy and/or one degenerate energy level and is different from the conventional electron or hole carriers that correspond to a continuous energy band .
the low mobility carriers in crystal 2 are hole type ( @xmath54 10 cm@xmath1/vs ) and they are an order of magnitude less mobile when compared to hole carriers in crystal 1 at 300 k. note that hole carriers in crystal 1 at room temperature constitute high mobility band ( @xmath56 10@xmath1 cm@xmath1/vs ) . yet , just like @xmath57 , the @xmath58 band exhibits sign change below 50 k , having identical temperature dependence when compared to @xmath57 with up to three orders of magnitude lower nominal values .
the @xmath59 carriers are electron type in crystal 2 .
they show little change in magnitude and do not change sign .
the most striking difference between crystals 1 and 2 is the ratio of nominal mobility values .
the large mobility difference is absent in crystal 2 .
moreover , there is about or less than one order of magnitude between @xmath59 and @xmath58 in the high thermopower region from 10 to 40 k. the low and high carrier concentrations in both crystals have nearly identical values , suggesting a compensated nature of electronic transport at all temperatures .
the carrier concentrations @xmath60 and @xmath61 are rather close at 300 k but they are both about an order of magnitude lower and have the same temperature dependence as carrier concentrations @xmath62 and @xmath63 in crystal 1 .
carrier concentrations in crystals 1 and 2 are nearly identical in the temperature region of high thermopower , whereas in the impurity regime below 10 k crystal 2 ( no mit ) shows much higher carrier concentrations .
values of carrier mobilities for crystal 2 ( @xmath310@xmath64 cm@xmath1/vs ) are in general agreement with mobility values at low temperatures obtained using a two - band model ( 4 - 25 k ) on crystals with no mit.@xcite using carrier concentrations and hall mobilities from a large number of measured @xmath45 isotherms we obtain _
s(t ) _ for both crystals in the framework of a two - band non - interacting semiconductor @xcite model where thermopower @xmath65 ) and @xmath66-\xi_{e , h}\ ] ] where @xmath67 .
the scattering exponent @xmath68 represents the energy dependence of the relaxation time @xmath69=@xmath70@xmath71 , @xmath72=@xmath73/(k@xmath74 t ) are the reduced fermi energies for electrons and holes @xmath75 and @xmath76 ( as measured from the bottom of the conduction band for electrons and from the top of the valence band for holes ) and @xmath73 = h@xmath1/(2m@xmath77)(n / v)@xmath78(3/8@xmath30)@xmath78 .
hence , for semimetals @xmath79 = - ( @xmath80 + @xmath81 and @xmath75= - @xmath76 - @xmath82/(k@xmath74 t ) where @xmath82 is the overlap energy , and for semiconductors @xmath79 = - @xmath80 + @xmath83 and @xmath75 = - @xmath76 - @xmath82/(k@xmath84 where @xmath83 is the energy gap for semiconductors . in this model
we used value _
s = -1/2 _ assuming that acoustic phonon scattering is dominant at all temperatures and @xmath85 = @xmath86 .
the two - band non - interacting semiconductor model adequately describes crystal 2 above 50 k and below 5 k , whereas it fails to explain @xmath87 for crystal 1 in its metallic state ( fig .
while the similar shape of calculated and measured thermopower for both crystals between 10 and 40 k in the noninteracting model argues in favor of the two - band approach , such calculation does not explain the amplification of @xmath87 observed in the ( 10 - 30 ) k range .
the prime suspect for discrepancy are strong correlations which can have a significant impact on carriers in _
3d _ bands . in a correlated electronic system with two bands separated by a gap ,
_ s _ depends not only on the gap size but also on the anisotropy or asymmetry in the transport function:@xcite @xmath88 where @xmath12@xmath13 is the chemical potential , @xmath89 is the gap and @xmath90@xmath23=(@xmath23@xmath40-@xmath23@xmath41)/(@xmath23@xmath40+@xmath23@xmath41 ) is the asymetry parameter .
the asymmetry parameter carries the information on band specific correlation strengths since @xmath23@xmath91=@xmath92m@xmath93e@xmath94/@xmath95@xmath91m@xmath96 where @xmath97 are quasiparticle weights , @xmath95@xmath91 are scattering amplitudes and @xmath98 is the effective mass of carriers in conduction and valence bands .
in addition , in real materials the @xmath99 rapidly diminishes with the increase in impurity concentration due to the increase of scattering amplitude , as recently observed in fesb@xmath0.@xcite bandwidth narrowings @xmath98 influence _
s(t ) _ not only through the asymmetry parameter @xmath100 but also via chemical potential since @xmath73=(3k@xmath74t/4)ln(m@xmath101.@xcite finally , the presence of impurities can have considerable effect on both @xmath99 and _ s(t)_. the _
s(t ) _ magnitude at a given temperature depends on the ability of impurity carriers to put the fermi level in the optimal position for thermopower enhancement.@xcite both crystals have similar _
s(t ) _ and nearly identical magnetothermopower _
mt = [ s(9t)-s(0)]/s(0 ) ] _ in the region of high _ s _ ( fig .
in contrast to differences in @xmath17 , thermopower _
s(t ) _ for heat flow along the _ c_-axis is rather similar above 120 k and below 8 k. thermopower changes sign from positive to negative above 100 k in both crystals , indicating the presence of two carrier types .
since @xmath98=e@xmath102/@xmath103 where @xmath102 is the scattering time and @xmath103 is the mobility in conduction and valence bands , we use mobility values for individual bands obtained from the fits of the @xmath104 in the two - band model . the value of @xmath105@xmath69@xmath106/@xmath95 then becomes a fit parameter for each carrier band , in addition to the chemical potential @xmath12@xmath13 .
fits to the above equation for a fixed value of the gap which corresponds to the correlated electron temperature region ( 10 - 30 ) k are shown in fig .
3(a ) as red ( crystal 1 ) and blue ( crystal 2 ) solid lines .
the best fits are obtained for the gap values of @xmath3 ( 15 - 20 ) mev ( table i ) , suggesting that the large enhancement of _ s _ is due to the strong electronic correlations associated with a smaller indirect gap.@xcite note that this is within the extremal limits in the assymetric case when the chemical potential is near the edge of the valence band ( @xmath107@xmath108 @xmath89/t+5/2k@xmath74).@xcite for large @xmath95 or for similar @xmath23@xmath40 and @xmath23@xmath41 ( symmetric multiband effects ) @xmath1090 .
this is observed : _
s _ sign change is at @xmath110 4 ) k , for both crystals 1 and 2 .
this implies that correlation effect on _
s(t ) _ vane at high temperatures .
the phonon drag is unlikely to have a significant contribution to _
s _ since isostructural rusb@xmath0 and feas@xmath0 have larger @xmath9 and much smaller values of @xmath87@xmath111 due to different temperature dependence of _ s _ and nernst coefficient.@xcite the @xmath112s(t)@xmath112 rises below 40 k , reaching values of 1 mv / k ( crystal 1 ) and 1.9 mv / k ( crystal 2 ) in the ( 10 - 20 ) k range ( fig .
4 ) . the @xmath113@xmath2 of crystal 1 ( 7800 @xmath4wk@xmath5@xmath6 )
is maximized at @xmath114 = 28 k , which is higher than in crystal 2 ( @xmath114 = 20 k ) and close to t@xmath115 .
the maximum value we find is three times larger than the tpf previously observed@xcite in fesb@xmath0 and occurs at 16 k higher temperature . for fesb@xmath0 crystals 1 ( red ) and 2 ( blue squares ) .
red and blue balls show fits to the two band noninteracting semiconductor model .
fits to the correlated electron model for crystal 1 are shown by red , purple and brown ( crystal 1 ) and dark blue , light blue and violet ( crystal 2 ) solid lines that correspond to increasing values of fixed energy gap @xmath89 ( see text ) .
the ratio of individual band mobilities in a two - carrier model is shown by the red / orange ( crystal 1 ) and dark / light blue squares ( crystal 2 ) . ]
the central finding of the above analysis is that large thermopower in both fesb@xmath0 crystals can be rationalized within the correlated electron model with two bands having the large disparity of quasiparticle properties .
this is best illustrated by the ratio of carrier mobilities in the conduction and valence bands . while the mobility of the valence band carriers in crystal 1 greatly exceeds that of the conduction carriers , they are practically immobilized in crystal 2 where conduction carriers have higher mobility .
a key observation here is that despite great disparity in absolute values , mobilities of the valence band carriers in two crystals show strikingly similar temperature dependence with a pronunced anomaly at mit .
most importantly , in both cases there is a change in the nature of valence charge carriers , which are holes at temperatures above mit and electrons at lower temperatures .
as we discuss later , this observation reveals the likely nature of the mit and of its absence in crystal 2 .
now we turn to the quasi - one - dimensional ( quasi-1d ) metallic conductivity@xcite and the mit in crystal 1 which is the key for colossal tpf . for
a large concentration of impurities the chemical potential can go into the valence / conduction band , producing metallic @xmath17.@xcite while an order of magnitude disparity in the number of charge carriers observed in crystals 1 and 2 above @xmath3100 k would be consistent with such a scenario , it can not be reconciled with the closely compensated nature of the carrier content . below 10 k , where conductivity is governed by impurities ,
there are more carriers in crystal 2 than in crystal 1 .
finally , impurity bands can not account for the quasi-1d metallic conductance .
band structure suggests that the likely origin of the quasi-1d transport is the non - bonding @xmath116 band , where the overlaps of fe @xmath116 orbitals are along the chains of edge - sharing octahedra parallel to the _ c _ axis , with little or no overlaps between orbitals in different chains [ fig .
5 ( a)].@xcite hall data shows that quasi-1d metallic conductance in sample 1 at temperatures above 40 k is provided by a small number of mobile holes , which implies a nearly filled valence band .
such a non - bonding band must be narrow and strongly correlated , and should be described by the 1d hubbard model.@xcite the low - dimensional transport in such band is sensitive to disorder and mit is expected at half filling
. can it be the origin of the mit observed in crystal 1 that @xmath116 band is depleted with the decreasing temperature , attaining half - filling at @xmath35 ?
could it then be that a small additional disorder in crystal 2 induces localization of the quasi-1d charge carriers ? ( a ) shows 2x2x2 _ pnnm _ unit cell as seen from the top of the _ c_-axis ( left ) , and from the top of the _ a_-axis ( right ) .
shaded rectangles denote short fe - sb distances within the fesb@xmath117 octahedra .
intrachain and interchain fe - sb - fe angles are denoted as @xmath118@xmath119 and @xmath118@xmath0 , respectively .
next - nearest - neighbour interchain fe - fe distance is indicated by the double arrow .
( b ) temperature dependence of @xmath120 and @xmath121 for crystal 1 ( red ) and crystal 2 ( blue ) .
( c ) adp factors for fe ( anisotropic ) and sb ( isotropic ) for crystal 1 ( red ) and crystal 2 ( blue ) . vertical dashed lines in ( b ) and ( c ) at 40k and 100k indicate temperatures of insulator - metal transition and change of the nature of the charge carriers , respectively . ] in order to answer this question , we performed the pdf analysis of the x - ray data taken on powder samples of crystals 1 and 2 ( figs .
5 and 6 ) .
the pdf provides insight into the crystal structure on short and intermediate length - scales and allows quantifying subtle structural features such as bond properties and local disorder .
intrachain and interchain fe - sb - fe angles , @xmath118@xmath119(t ) and @xmath118@xmath0(t ) respectively , shown in fig .
5(b ) , are very similar , with a small difference developing in @xmath118@xmath119 above 100 k. this may indicate charge redistribution involving @xmath116 bands since the presence of fe @xmath116 charge will result in a somewhat larger @xmath120 angle.@xcite atomic displacement factors ( adp ) [ fig .
5(c ) ] of fe atoms are rather anisotropic : mean - square atomic displacements along the _ a _ and the _ b _ axes , @xmath122 and @xmath123 , are nearly equal for both crystals , while those along the _ c _ axis , @xmath124 , are noticeably enhanced in crystal 2 at all temperatures , being 0.0084 @xmath125 ( crystal 2 ) and 0.0054 @xmath125 ( crystal 1 ) at 10 k. enhanced adps are typical indication of disorder . in this case disorder is markedly 1d , precisely along the _ c _ axis and significantly higher in crystal 2 .
it is clear from the difference curves ( fig .
6 ) in the x - ray rdf ( refs . 23 and 24 ) that the average local environments on the length scale of about one _ pnnm _ unit cell ( reflected in the first several rdf peaks up to 4.5 @xmath24 ) are indistinguishable
. on larger length scales structural features start to differ .
the rdf peak at about 4.65 @xmath24 , denoted by a vertical arrow in fig .
6 , is markedly broader and less intense for crystal 2 at all temperatures studied .
this peak corresponds to the next - nearest - neighbor fe - fe distance [ fesb@xmath117 interchain distance , double arrow in figure 5(a ) ] , with no other contributions .
such a discrepancy could have several origins arising in a difference in fe - fe bondlength distribution , amount of fe in the structures of the two crystals , charge state of fe between the two crystals , or a combination of these effects .
low - dimensional @xmath8 is very sensitive to disorder which , in the simplest picture , suppresses metallic state by inducing strong localization . more generally , the disorder can impact the @xmath116 overlap and the band structure , the orbital character of the electronic states responsible for conduction , the occupancy of fe @xmath116 orbitals , and even the orbital - dependent hubbard _ u _ interaction strength in a @xmath116 quasi-1d band of itinerant states that forms with sufficient @xmath116 overlap .
as opposed to this `` self - generated '' impurity level arising from fe _ d _ orbitals in crystals with mit , extrinsic impurity levels can not produce metallic resistivity near or close to room temperature as in crystals with mit and are significant only at low temperatures.@xcite crystals with mit were cooled slowly from 1000 @xmath19c to 650 @xmath19c and were decanted at that temperature from the liquid sb flux . on the other hand , crystals with no mit were cooled to 640@xmath19c , closer to sb solidification temperature and were subsequently decanted at 690@xmath19c .
there are two main factors that could contribute to increased crystallographic disorder in crystals with no mit .
fesb@xmath0 melts and decomposes at 738@xmath19c to fesb and sb .
hence , crystallization occurs over a rather narrow temperature window .
crystals decanted at 690@xmath19c are much closer to the melting point when compared to crystals decanted at 650@xmath19c .
this could contribute to increased disorder in fe - sb chemical bonds .
in addition , crystals cooled to 640 @xmath19c are closer to solidification line and are likely to experience more stress from the flux .
this is consistent with observed structural differences , suggesting that mit in fesb@xmath0 crystals is governed by subtle structural differences tunable by synthesis procedure .
greatly increased conductivity near mit , combined with significant electronic tep , leads to a new record high thermoelectric power factor .
in summary , we report the highest known tpf induced by the synthesis - controlled mit in the correlated electron semiconductor fesb@xmath0 . the large thermopower enhancement can be understood within the electronic model , whereas mit likely originates in quasi-1d and strongly correlated narrow band of itinerant states very sensitive to disorder .
this is further supported by the recent observation of quasi-1d magnetism in isostructural crsb@xmath0.@xcite
we thank t. m. rice and simon billinge for useful discussion and milinda abeykoon and pavol juhas for help with x - ray experiments . this work was carried out at the brookhaven national laboratory , which is operated for the u.s .
department of energy by brookhaven science associates de - ac02 - 98ch10886 .
the advanced photon source at argonne national laboratory , operated by uchicago argonne llc , is supported under the u.s .
doe - os contract no .
de - ac02 - 06ch11357 .
neutron pdf experiments were carried out on hipd at lansce , funded by doe bes ; lanl is operated by los alamos national security llc under de - ac52 - 06na25396 . | we show that synthesis - induced metal -insulator transition ( mit ) for electronic transport along the orthorombic _
c _ axis of fesb@xmath0 single crystals has greatly enhanced electrical conductivity while keeping the thermopower at a relatively high level . by this means
, the thermoelectric power factor is enhanced to a new record high s@xmath1@xmath2 @xmath3 8000 @xmath4wk@xmath5@xmath6 at 28 k. we find that the large thermopower in fesb@xmath0 can be rationalized within the correlated electron model with two bands having large quasiparaticle disparity , whereas mit is induced by subtle structural differences .
the results in this work testify that correlated electrons can produce extreme power factor values . |
in an introductory course on quantum theory one commonly discusses some of the simplest models , such as , for example : free particle , particle in a box , harmonic oscillator , tunnelling through a square barrier , etc . with the purpose of making the students more familiar with the principles or postulates of quantum theory .
some time ago gibbs@xcite introduced the quantum bouncer as a model for the pedagogical discussion of some of the relevant features of quantum theory .
he derived the solutions to the schrdinger equation in terms of airy functions and obtained the energy spectrum for two different model settings .
a closely related model , the harmonic oscillator with a hard wall on one side , had been discussed earlier by dean@xcite and later by mei and lee@xcite .
this model is as simple as the quantum bouncer and both can therefore be discussed in the same course .
an interesting feature of this model is that it may in principle be useful to simulate a particle attached to a wall like , for example , an atom adsorbed on a solid surface@xcite .
although it is an oversimplified one dimensional model of the actual physical phenomenon we deem it worthwhile to discuss some of its properties in this paper . in sec .
[ sec : model ] we introduce the model , write the schrdinger equation in dimensionless form , and discuss some of the properties of its solutions . in sec .
[ sec : results ] we obtain the eigenfunctions and eigenvalues explicitly in terms of the well known weber functions@xcite and show the behavior of the eigenvalues and excitation energies with respect to the distance between the particle and the wall .
we also calculate the zero point energy for values of the model parameters corresponding to de adsorption of h on pd(100)@xcite and outline the application of the rayleigh
ritz variational method to the schrdinger equation . finally , is sec .
[ sec : comments ] we give further reasons why this model may be useful in a course on quantum theory .
we consider a simple one dimensional model for a particle of mass @xmath0 attached to a surface located at @xmath1 that separates free space ( @xmath2 ) from the bulk of the material ( @xmath3 ) .
therefore , we assume that the potential exhibits an attractive tail for @xmath2 that keeps the particle in the neighborhood of the surface and a repulsive one for @xmath3 that prevents the particle from penetrating too deep into the material . since we want to keep the model as simple as possible we choose @xmath4 where @xmath5 .
we note that the particle oscillates about @xmath6 but the motion is not harmonic because of the effect of the hard wall .
the schrdinger equation reads @xmath7 where the boundary condition at @xmath1 comes from the fact that in this simple model the particle can not penetrate into the material and , consequently , @xmath8 for all @xmath3 .
since it is more convenient to work with a dimensionless equation we define the length unit @xmath9 and the dimensionless coordinate @xmath10 .
thus , the dimensionless schrdinger equation reads @xmath11 where @xmath12 it is worth noting that @xmath13 increases with @xmath14 , @xmath0 , and @xmath15 .
it is precisely eq .
( [ eq : schr_dim ] ) that was discussed by dean@xcite and mei and lee@xcite . when @xmath16 we have the well known harmonic oscillator with eigenvalues @xmath17 on the other hand , when @xmath18 we have the harmonic oscillator in the half line and @xmath19 note that these are merely the harmonic oscillator eigenvalues with odd quantum number ( the corresponding eigenfunctions have a node at origin ) .
it follows from eq .
( [ eq : de / db_2 ] ) in the appendix that @xmath20 from which we conclude that the energy eigenvalues decrease monotonously between the following limits : @xmath21 when @xmath22 .
it should be kept in mind that the number of zeros of @xmath23 remains unchanged as @xmath13 goes from @xmath24 to @xmath25 .
equation ( [ eq : de / db_2 ] ) is useful for obtaining the asymptotic behavior of the energy as @xmath16 . to this end
we simply integrate it from @xmath26 to @xmath27 : @xmath28 for the ground state we expect that @xmath29 as @xmath16 .
the normalization factor is approximately given by @xmath30\ ] ] where @xmath31 is the error function .
since @xmath32 we write @xmath33 and expand @xmath34 in a taylor series about @xmath35 : @xmath36 finally , eq . ( [ eq : e(-q0 ) ] ) yields @xmath37 in order to obtain this result we substituted @xmath38 after the integration and neglected a term proportional to @xmath39 because it is much smaller than the exponential one retained in eq .
( [ eq : e0(-q0 ) ] ) .
this result agrees with the one derived by mei and lee@xcite by means of perturbation theory ( note that their parameter @xmath40 is @xmath41 ) .
proceeding in the same way for the first excited state we obtain @xmath42 that is slightly different from the result of mei and lee@xcite .
however , they are equivalent for most purposes because the difference between them is smaller than their absolute errors .
( [ eq : virial_2 ] ) gives us the virial theorem@xcite for this model @xmath43 where @xmath44 .
the right hand side of this equation is the virial of the force exerted by the surface .
( [ eq : hv2 ] ) provides us with another interesting relation @xmath45 that clearly reveals the asymmetry of the interaction between the particle and the surface . in both cases
we recover the well known results for the harmonic oscillator @xmath46 and @xmath47 when @xmath16 .
if we define the new independent variable @xmath48 and write the energy as @xmath49 then we realize that @xmath50 is a solution to the weber equation@xcite @xmath51 the general solution is@xcite
@xmath52 \label{eq : weber_func}\ ] ] where the confluent hypergeometric function @xmath53 is a solution to @xmath54 and can be expanded in a taylor series about @xmath55 as @xmath56 the boundary condition @xmath57 enables us to calculate the eigenvalues from the roots of @xmath58 . for each value of @xmath13
we solve @xmath59 for @xmath0 and then calculate the dimensionless energy @xmath49 .
this approach has already been discussed by dean@xcite and mei and lee@xcite .
[ fig : en ] shows @xmath60 for @xmath61 and a wide range of values of @xmath13 .
we note that the dimensionless energy decreases monotonously as predicted by eq .
( [ eq : de / dq0 ] ) between the limits indicated in eq .
( [ eq : en_bounds ] ) .
the gap between two consecutive energy levels of the harmonic oscillator is @xmath62 ( @xmath63 ) .
[ fig : de ] shows @xmath64 for @xmath65 where we note that the energy gap increases with @xmath66 revealing that the presence of the wall results in an anharmonic oscillation .
it becomes more harmonic as @xmath13 increases ( by increasing either @xmath14 , @xmath0 or @xmath15 ) . in order to have a clearer physical idea of the kind of predictions of this simple model we may choose the parameters for the adsorption of hydrogen on pd(100 ) .
gladys et al@xcite estimated @xmath67 and @xmath68 for hydrogen that lead to @xmath69 .
the zero point energy for such model parameters is approximately @xmath70 instead of the value @xmath71 chosen by those authors .
although they fitted the potential energy of the vertical displacement of the h atom from the pd surface to a cubic polynomial they simply chose the zero point energy of the harmonic oscillator .
present model predicts that the zero point energy is slightly greater than the harmonic
oscillator one because of the repulsive effect of the surface .
we agree that a hard wall may not be the most adequate representation of the short range interaction between the h atom and the pd surface , but we think that the model is a reasonably simple first approach to the physical phenomenon .
for deuterium we have @xmath72 which , together with the same force constant and about twice the mass , yields @xmath73 and @xmath74 that is closer to the harmonic oscillator zero point energy@xcite .
we note the effect of the distance to the surface and the mass of the particle on the vibrational energies .
this model is also useful for discussing the rayleigh
ritz variational method@xcite .
if , for example , we choose the trial function @xmath75 as a linear combination of the non orthogonal basis set @xmath76 and minimize the approximate energy @xmath77 then we arrive at the secular equations @xmath78 there are nontrivial solutions for the coefficients @xmath79 only if the secular determinant vanishes @xmath80 its roots @xmath81}$ ] , @xmath82 are upper bounds to the actual eigenvalues and satisfy @xmath83}>w_{n}^{[n+1]}>\epsilon _
{ n } \label{eq : var_bounds}\ ] ] the particle attached to a wall is therefore a suitable example for illustrating how the roots of the secular determinant approach the eigenvalues ( calculated accurately from the weber functions ) from above as @xmath84 increases .
we do not show results here and simply mention that the calculation is greatly facilitated by the fact that one calculates the integrals @xmath85 and @xmath86 analytically .
the model discussed here is suitable for a course on quantum theory because it does not require much more mathematical background than it is necessary for the discussion of the well known harmonic oscillator or the quantum bouncer@xcite .
it is useful for introducing a numerical calculation of the eigenvalues that the student does not find in the treatment of the harmonic oscillator .
one can approach the problem by means of either the weber functions or the rayleigh
ritz variational method .
the student will also learn that it is necessary to modify the form of the well known virial theorem and other mathematical expressions in order to take into account the effect of the wall .
we believe that present derivation of the analytical results in sec .
[ sec : model ] and in the appendix is simpler than those available in the scientific literature@xcite .
in addition to it , the model enables us to simulate the adsorption of an atom on a surface and discuss anharmonic vibrations in quantum theory . in the study of molecular vibrations
one introduces nonlinear oscillations by means of cubic , quartic , and other terms of greater degree in the potential energy function . in this case
it arises from the boundary condition forced by the hard wall .
in this appendix we develop some useful analytical results for the eigenfunctions and eigenvalues of the constrained oscillator .
although similar expressions have already been shown elsewhere@xcite we derive them here in a form that is more suitable for our needs .
consider the dimensionless schrdinger equation @xmath87 both the eigenvalue @xmath88 and the eigenfunction @xmath89 depend on the chosen value of @xmath90 .
if we differentiate this equation with respect to @xmath90 and call @xmath91 we have @xmath92 if we multiply this equation by @xmath89 and integrate the result between @xmath90 and @xmath25 we easily obtain @xmath93 because the integration by parts of the first term yields @xmath94 \chi ( x)\varphi ( x)\,dx\ ] ] we define the operators @xmath98 and @xmath99 , where @xmath100 .
the commutator between them reads @xmath101=\hat{h}\hat{v}-\hat{v}\hat{h}=-\hat{d}^{2}-xv^{\prime } \ ] ] if @xmath89 is an eigenfunction of @xmath102 with eigenvalue @xmath88 then straightforward integration by parts leads to @xmath103\varphi \,dx=-\frac{1}{2}b\varphi ^{\prime } ( b)^{2 } \label{eq : virial_1}\ ] ] which , by virtue of eq .
( [ eq : de / db_2 ] ) , becomes the virial theorem @xmath104 where @xmath105 we can test these equations quite easily by means of the well known solutions to the free harmonic oscillator @xmath109 where @xmath110 is a hermite polynomial and @xmath111 the corresponding normalization factor@xcite .
if we choose @xmath90 to be one of the zeroes @xmath112 , @xmath113 of @xmath110 , @xmath114 then we verify that @xmath115 satisfies equations ( [ eq : virial_1 ] ) and ( [ eq : hv1 ] ) for @xmath116 .
if , for example , @xmath117 for @xmath118 and @xmath119 for @xmath120 then @xmath121 is the energy of the ground state of the harmonic oscillator with the boundary condition at @xmath117 and @xmath122 is the energy of the ground state with @xmath123 and of the first excited state with @xmath124 . | we present a simple one dimensional quantum mechanical model for a particle attached to a surface . we solve the schrdinger equation in terms of weber functions and discuss the behavior of the eigenvalues and eigenfunctions .
we derive the virial theorem and other exact relationships as well as the asymptotic behaviour of the eigenvalues .
we calculate the zero point energy for model parameters corresponding to h adsorbed on pd(100 ) and also outline the application of the rayleigh ritz variational method . |
this paper is about the gradient flows on compact surfaces , thus the reference to abbott s _ flatland _ @xcite in the title .
the paper is an informal introduction into the philosophy and some key results from @xcite -@xcite , as they manifest themselves in @xmath0 . the remarkable convergence of topological , geometrical , and analytical approaches to the study of closed surfaces is widely recognized by the practitioners for more than a century .
we will exhibit a similar convergence of different investigative approaches to vector flows on surfaces _ with boundary_. we will take advantage of the relative simplicity of @xmath0 flows to illustrate and popularize the main ideas of our recent research of _ traversally generic
_ flows on manifolds with boundary .
when the results are specific to the dimension two , their validation will be presented in detail .
the multidimensional arguments that resist significant simplifications in @xmath0 will be described and explained in general terms . throughout the investigation
, we focus on the interactions of gradient flows with the boundary , rather than on the critical points of morse functions .
so , in our approach to the morse theory , the boundary effects rule .
_ morse theory _
, the classical book of john w. milnor @xcite , starts with the canonical picture of a morse function @xmath1 on a 2-dimensional torus @xmath2 ( see fig .
it is portrayed as the height function @xmath3 on the torus @xmath2 residing in the space @xmath4 .
the height @xmath3 has four _ critical points _ : @xmath5 , and @xmath6 so that @xmath7 a point @xmath8 is called critical if the differential @xmath9 of @xmath3 vanishes at @xmath8 . in the vicinity of each critical point @xmath8 , @xmath2 admits a pair of local coordinate functions , say @xmath10 and @xmath11 , so that locally the function @xmath3 acquires the form @xmath12 where the signs may form four possible combinations .
[ fig1.1 ] we call a vector field @xmath13 , tangent to @xmath2 , _ gradient - like _ if @xmath14 everywhere outside of the set @xmath15 of critical points .
if the torus is slightly slanted " with respect to the vertical coordinate @xmath3 in @xmath4 , then the following picture emerges .
the majority of downward trajectories of the @xmath3-gradient flow @xmath16 that emanate from @xmath17 , asymptotically reach @xmath6 .
there are two trajectories that asymptotically link @xmath17 with @xmath18 , and two trajectories that link @xmath17 with @xmath19 .
no ( unbroken ) trajectory asymptotically connects @xmath18 to @xmath19 .
perhaps , a more transparent depiction of the gradient flow @xmath16 is given in fig .
2 , where the torus is shown in terms of its fundamental domain , the square . to form @xmath2 ,
the opposite sides of the square are identified in pairs .
the morse theory is concerned with the sets of _ constant level _ @xmath20 and the _ below constant level _ sets @xmath21 .
the main observation is that the topology of these sets is changing in an essential way only when the rising @xmath22 crosses the critical values @xmath23 each such critical crossing " results in an _ elementary surgery _ on the set @xmath21 , where @xmath22 is just below a critical value @xmath24 . for a small @xmath25 , an elementary surgery @xmath26 attaches the _ handle _
@xmath27 to the set @xmath28 .
eventually , when @xmath22 rises above @xmath29 , the entire topology of torus @xmath2 is captured by a sequence of these elementary surgeries . from a different angle ,
the knowledge of how the critical points @xmath30 interact via the trajectories of the @xmath31-flow is also sufficient for reconstructing the surface @xmath2 as fig .
2 suggests ( see @xcite ) .
[ fig1.2 ] note that , in the vicinity of each critical point , the gradient flow exhibits _ discontinuity _ : small changes in the initial position of a point @xmath8 , residing in the vicinity of a critical point , result in significant differences in the position of @xmath32 for big positive / small negative values of @xmath33 ( see fig .
3 ) . in fact , this discontinuity of the gradient flow , expressed in terms of the _ stable _ and _ unstable manifolds _ of critical points ( see @xcite ) , captures the topology of the surface ( as the left diagram in fig . 2 suggests ) ! [ fig1.3 ] as a result of gradient flow discontinuity , the _ space of trajectories _
@xmath34 is pathological ( non - separable ) .
the space @xmath34 is constructed by declaring equivalent any two points that reside on the same trajectory .
when a compact connected surface @xmath35 has a nonempty boundary @xmath36 , traditionally , the morse function @xmath37 is assumed to be constant on @xmath36 and its gradient flow interacts with the boundary in constrained way .
then the _ relative topology _ of the pair @xmath38 can be captured in the ways analogous to the previous description of the morse theory on torus .
in fact , the morse theory on manifolds with boundary can be viewed as a very special instance of the morse theory on stratified spaces ( the two strata @xmath36 and @xmath35 form the stratification ) .
the latter was developed by goresky and macpherson in [ gm ] -[gm2 ] . in this paper
, we propose a different philosophy for the morse theory on compact surfaces / manifolds @xmath35 _ with boundary_. to formulate it , let us revisit our favorite closed surface , the torus . by deleting from @xmath2 small disks , centered on the points of the critical set @xmath15 , we manufacture a surface @xmath35 whose boundary is a disjoint union of four circles . evidently , @xmath37 has no critical points at all .
still it has a nontrivial topology ! can this topology be reconstructed from some data , provided by the critical point - free @xmath3 and its gradient - like field @xmath39 ?
an experienced reader would notice that the restriction @xmath40 has critical points ( maxima and minima ) , some of which interact _ along _ the boundary ( with the help of a gradient - like field @xmath41 , tangent to @xmath36 ) .
however , it is quite clear that these interactions are not sufficient for a reconstruction of the topology of @xmath35 ! in fact , a reconstruction of the surface @xmath35 becomes possible if one introduces additional interactions between the points of @xmath42 that occur through the bulk @xmath35 " and are defined with the help of _ both _ vector fields @xmath13 and @xmath41 .
this observation has been explored by a number of authors , but it is not the world view that we are promoting here ... to dramatize further the situation we are facing , let us place four small disks , centered on the critical points of @xmath1 , into a single open disk @xmath43 and form @xmath44 ( see fig 2 , the right diagram ) .
again , @xmath45 has no critical points , the gradient field @xmath46 , but its topology of @xmath35 is nontrivial .
this time , the boundary @xmath36 of the punctured torus @xmath35 is just a single circle !
let us keep this challenge in mind .
can one propose a morse theory " that is not centered on critical points ?
the answer is affirmative .
it relies on the following observation .
typically , in the vicinity of @xmath36 , the @xmath13-trajectories are interacting with the boundary in a number of very particular and stable ways : they are either transversal to @xmath36 , or are tangent to it in a _ concave _ or _ convex _ fashion for which some trajectories will be cubically tangent to the boundary , but the majority of vector fields @xmath13 avoid such cubic tangencies . ]
( see fig .
so the boundary @xmath35 may be wiggly " with respect to the flow .
we claim that this geometry of the @xmath13-flow in connection to the boundary @xmath36 is the crucial ingredient for reconstructions of @xmath35 in terms of the flow ( see section 8 , especially theorem [ th7.1 ] ) . in the vicinity of a
_ concave _ tangency point , the @xmath13-flow is _ discontinuous _ in the same sense as the gradient flow is discontinuous in the vicinity of its critical point : in time , close initial points become distant . in this context
, the divergence of initially close points occurs due to very different _ travel times _ available to them ; unlike the infinite travel time for the gradient flows of the morse theory on _ closed _ surfaces , in the case of the non - singular gradient flows on surfaces with boundary , every point exits the surface in finite time .
in particular , the surface is _ not _ flow - invariant . and
again , these discontinuities of the flow reflect the topology of the surface .
let us clarify this point .
4 shows a gradient flow @xmath13 on a surface @xmath47 , the disk with @xmath48 holes .
the nonsingular function @xmath37 is the vertical coordinate in @xmath49 .
each @xmath13-trajectory is either a closed segment , or a singleton . by collapsing each trajectory to a point
, we create a quotient space @xmath34 of trajectories . since the flow trajectories are closed segments or singletons , this time , the trajectory space @xmath34 is decent " , a finite graph with verticies of valency @xmath50 or @xmath51 only .
the verticies of valency @xmath51 correspond to the points on @xmath36 where the boundary is concave with respect to the flow , and the univalent verticies to the points on @xmath36 where the flow is convex .
the obvious map @xmath52 cellular .
moreover , because the fibers of @xmath53 are contractable , @xmath53 is a _
homotopy equivalence_. in particular , the fundamental groups @xmath54 and @xmath55 are isomorphic with the help of @xmath53 . so the trajectory spaces of generic non - vanishing vector fields @xmath13 of the gradient type on connected surfaces @xmath35 with boundary deliver @xmath50-dimensional _ homotopy theoretical models _ of @xmath35 . [ fig1.4 ]
following @xcite , for any vector field @xmath13 on a compact surface @xmath35 with boundary such that @xmath56 , we consider the closed locus @xmath57 , where the field is pointing inside @xmath35 and the closed locus @xmath58 , where it points outside .
the intersection @xmath59 is the locus where @xmath13 is _ tangent _ to the boundary @xmath36 .
points @xmath60 come in two flavors : by definition , @xmath61 when @xmath62 points inside of the locus @xmath57 ; otherwise @xmath63 .
to achieve some uniformity of notations , put @xmath64 and @xmath65 .
[ fig1.5 ] we say that a vector field @xmath13 on a compact surface @xmath35 is _ boundary generic _ if : * @xmath66 , viewed as a section of the normal @xmath50-dimensional ( quotient ) bundle @xmath67 is transversal to its zero section , * @xmath68 , viewed as a section of the normal @xmath50-dimensional bundle @xmath69 , is transversal to its zero section .
@xmath70 in particular , for a boundary generic @xmath13 , the loci @xmath71 are finite unions of closed intervals and circles , residing in @xmath36 ; and the loci @xmath72 are finite unions of points , residing in @xmath36 ( see fig .
we denote by @xmath73 the space ( in the @xmath74-topology ) of all boundary generic fields on a compact surface @xmath35 .
let @xmath75 denote the euler number of a space @xmath76 .
recall that @xmath75 is the alternating sum of dimensions of the homology spaces @xmath77 . since for a connected surface @xmath35 with boundary @xmath78
, we get @xmath79 for a closed connected surface , @xmath80 given a vector field @xmath13 with isolated zeros , we can associate an integer @xmath81 with each zero @xmath10 of @xmath13 .
this integer is the _ degree _ of the map which , crudely speaking , takes each point @xmath8 on a small circle @xmath82 with its center at @xmath10 to the unit vector @xmath83 .
then we define @xmath84 , the ( global ) index of @xmath13 , as the sum @xmath85 .
the morse formula @xcite , in the center of our investigation , computes the index @xmath84 of a given boundary generic vector field @xmath13 on a surface @xmath35 as the alternating sum of the euler numbers of the morse strata @xmath86 : @xmath87 in the case of a connected surface @xmath35 with boundary , @xmath88 , and this formula reduces to @xmath89 @xmath90 in particular , if @xmath39 , then @xmath91 , and we get @xmath92 where the rhs of the equation is the topological invariant @xmath93 of @xmath35 . in contrast , the cardinality @xmath94 depends on @xmath13 .
[ lem3.1 ] let a surface @xmath35 be formed by removing @xmath95 open disks from a closed surface @xmath96 , the sphere with @xmath97 handles .
then , for any boundary generic field @xmath39 on @xmath35 , @xmath98 moreover , @xmath99 only when @xmath100 .
the euler number is additive under gluing surfaces along their boundary components .
therefore , if @xmath95 disks are removed from @xmath96 , the sphere with @xmath97 handles , then @xmath101 .
thus the morse formulas ( [ eq1.1 ] ) and ( [ eq1.2 ] ) imply @xmath102 for any @xmath39 .
moreover , @xmath99 if and only if @xmath100 , the main feature of the _ boundary concave _ fields ( see definition [ def1.2 ] ) . in particular , for any non - vanishing boundary generic field @xmath13 on a torus with a single hole , @xmath103 ( cf .
2 ) . recall that an _ immersion _ is a smooth map of manifolds , whose differential has the trivial kernel .
consider a smooth map @xmath104 , which is an immersion in the vicinity of @xmath36 .
any such @xmath22 gives rise to the gauss map @xmath105 , defined by the formula @xmath106 , where @xmath107 is the tangent vector to @xmath36 at @xmath10 .
the direction of @xmath107 is consistent with the preferred orientation of @xmath36 , induced by the preferred orientation of @xmath35 .
let @xmath108 be a constant field on @xmath49 .
since the kernel of the differential of @xmath109 is trivial along @xmath36 , the field @xmath110 defines a vector field @xmath111 on @xmath35 in the vicinity of @xmath36 .
the pull - back field @xmath112 extends to a vector field @xmath13 on @xmath35 , possibly with zeros ( see [ g ] for engaging discussions of vector field transfers and the gauss - bonnet theorem ) .
then the degree of the gauss map is given by a classical hopf formula ( @xcite ) @xmath113 when @xmath104 is an immersion everywhere , the pull - back field @xmath114 everywhere .
thus @xmath91 , and , for a connected @xmath35 with @xmath115 , we get @xmath116 so , for an immersions @xmath22 , we get a new interpretation of formula ( [ eq1.2 ] ) : @xmath117 this global - to - local formula has another classical geometrical interpretation .
let @xmath118 be the riemannian metric on @xmath35 , the pull - back of the euclidean metric on @xmath49 .
let @xmath119 denote the normal curvature of @xmath36 with respect to @xmath120 .
then @xmath121 which leads to another pleasing global - to - local connection : @xmath122 in particular , for a connected orientable surface @xmath35 of genus @xmath97 with a _ single _ boundary component , @xmath123 so the number of @xmath13-trajectories @xmath124 in @xmath35 that are tangent to @xmath36 , but are not singletons ( they correspond to points of @xmath125 ) , as a function of genus @xmath97 , grows at least as fast as @xmath126 ! on the other hand ,
when @xmath36 is connected , by the whitney index formula @xcite , the degree of the gauss map @xmath105 can be also calculated as @xmath127 , where @xmath128 denotes the number of positive / negative self - intersections of the curve @xmath129 , and @xmath130 .
here is a brief description of the rule by which the self - intersections acquire polarities .
let @xmath131 be a point where the coordinate function @xmath132 attends its minimum on the curve @xmath133 .
if the tangent vector @xmath134 at @xmath135 , which defines the orientation of @xmath133 , is @xmath136 , then we put @xmath137 ; if @xmath138 , then @xmath139 .
starting at @xmath135 and moving in the direction of @xmath134 , we visit each self - intersection @xmath17 twice and in a particular order .
the first visitation defines a tangent vector @xmath140 , the second visitation defines a tangent vector @xmath141 .
when the ordered pair @xmath142 defines the clockwise orientation of the @xmath143-plane , then we attach @xmath144 " to @xmath17 .
otherwise , the polarity of @xmath17 is @xmath145 " .
therefore we get a somewhat mysterious connection between the self - intersections of @xmath36 under immersions @xmath104 and the tangency patterns of the flows in @xmath35 that are the @xmath22-pull - backs of non - vanishing flows in the plane .
[ th1.2 ] let @xmath108 be a vector field in the plane @xmath49 .
let @xmath35 be a connected orientable surface with a connected boundary .
consider an immersion @xmath104 such that the loop @xmath133 has transversal self - intersections only .
assume that the pull - back @xmath146 is a boundary generic field on @xmath35 .
then @xmath147 @xmath148 the latter inequality being sharp by an appropriate choice of @xmath22 .
the first formula is the result of combining the whitney formula for @xmath149 with formulas ( [ eq1.3 ] ) , ( [ eq1.4 ] ) . by a theorem of guth @xcite , for any immersion @xmath104
, the total number of self - intersections of the loop @xmath133 admits an estimate @xmath150 moreover , this lower bound is realized by an immersion @xmath104 !
therefore , by formula ( [ eq1.4 ] ) , the guth inequality is transformed into @xmath151 moreover , for some optimal immersion @xmath22 , @xmath152 when a surface @xmath35 is oriented and a field @xmath13 is boundary generic , then the points from @xmath125 come in two _ new flavors _ :
@xmath153 " . by definition ,
a point @xmath154 has the polarity
@xmath155 " if the orientation of @xmath156 determined by the pair @xmath157 , where @xmath158 is the inner normal to @xmath36 , agrees with the preferred orientation of @xmath35 .
otherwise , the polarity of @xmath17 is defined to be @xmath159 " .
thus , for each choice of orientation of @xmath35 ( and hence of @xmath36 ) we get a partition @xmath160 switching the orientation of @xmath35 switches the second polarities in the partition .
[ def1.2 ] we say that a boundary generic vector field @xmath13 is _ boundary convex _ if @xmath161 .
we say that a boundary generic @xmath13 is _ boundary concave _ if @xmath162 ( see fig 5 ) .
@xmath70 the existence of a boundary convex field puts severe restrictions on the topology of the surface .
[ lem1.2 ] if a compact connected surface @xmath35 with boundary @xmath115 admits a boundary convex gradient - like vector field @xmath39 , then @xmath35 is either a disk @xmath43 , or an annulus @xmath163 .
the convexity of the field @xmath13 implies that @xmath35 admits a @xmath164-directed continuous retraction on the locus @xmath57 .
since @xmath35 is connected , it follows that @xmath57 is connected as well .
thus , @xmath57 is either a circle , or a segment . in the first case , @xmath35 is diffeomorphic to an annulus @xmath165 $ ] ; in the second case , @xmath35 is diffeomorphic to a disk @xmath43 .
the same phenomenon occurs in any dimension : if a compact connected smooth @xmath166-manifold @xmath35 with a _ connected _ boundary admits a _
boundary convex _ gradient - like vector field @xmath39 , then @xmath167 ( @xcite ) . in other words
, @xmath168 is a topological obstruction to the existence of a boundary convex non - vanishing gradient field on @xmath35 .
in contrast , the boundary concave non - vanishing gradient fields are plentiful . for example
, consider a radial vector field @xmath13 on an annulus @xmath163 .
delete from @xmath163 any number of convex disks and restrict @xmath13 to the resulting @xmath169-disk with holes .
the convexity of the disks that we have removed implies that any disk with holes admits a boundary concave gradient - like vector @xmath39 .
many other surfaces admit such concave fields as well .
for example , consider a morse function @xmath170 on a closed surface @xmath96 and its gradient field @xmath13 .
then removing small convex ( in the local morse coordinates ) balls , centered on the critical points , from @xmath96 , produces a _
boundary concave _ non - vanishing gradient field on @xmath35 . in particular
if @xmath96 is a sphere with @xmath97 handles , then one can find a morse function with @xmath171 critical points ( see fig .
so the surface @xmath35 , obtained from @xmath96 by removing @xmath171 balls , admits a concave gradient - like field @xmath39 .
in fact , by theorem [ th6.2 ] , any connected orientable surface with boundary , but the disk , admits a boundary concave non - vanishing gradient field !
we view the integer @xmath172 as a _ measure of complexity _ of the @xmath13-flow , subject to the condition @xmath91 or , alternatively , subject to the condition @xmath39 .
we define the _ complexity _ of a compact connected surface @xmath35 with boundary as the minimum @xmath173 where @xmath13 runs over all non - vanishing boundary generic fields on @xmath35 . by varying @xmath13 within different spaces of fields , one may consider a variety of such minima ; non - vanishing fields and non - vanishing gradient - like fields are the two most important cases .
so we introduce the _ gradient complexity _
@xmath174 where @xmath13 runs over all non - vanishing gradient - like fields on @xmath35 .
evidently @xmath175 .
let @xmath176 denote the mbius band . in section 6
, we will show that @xmath177 , while @xmath178 , so the two notions of complexity are different . in terms of this complexity , we can restate the lemma 3.1 as follows .
[ cor 4.1 ] let @xmath35 be a connected compact surface with boundary .
let @xmath13 be a boundary generic vector field on @xmath35 , subject to the condition @xmath91 .
then the complexity of @xmath13 satisfies the inequality @xmath179 when @xmath180 , this inequality turns into the equality @xmath181 if and only if @xmath13 is boundary concave . as a result , for any natural @xmath182 , there are finitely many connected compact surfaces of bounded complexity @xmath183 .
in fact , the number of such surfaces ( counted up to a homeomorphism ) grows as a quadratic function in @xmath182 .
@xmath70 * example 4.1 .
* for any non - vanishing boundary concave field @xmath13 on the torus with a single hole , @xmath184 .
in fact , the constant field @xmath13 , being restricted to the complement to a convex disk in @xmath2 , is boundary concave and has the property @xmath184 .
thus , by corollary 4.1 , @xmath185 .
@xmath70 lemma [ lem3.1 ] leads immediately to [ cor4.2 ] let @xmath35 be a sphere with @xmath97 handles and @xmath95 holes , where @xmath186 . if @xmath35 admits a non - vanishing boundary concave field @xmath13 , then @xmath99 .
@xmath70 given a compact surface @xmath35 with boundary , we form its _ double _ @xmath187 by attaching two copies of @xmath35 along their boundaries .
note that @xmath188
. therefore , @xmath189 if and only if @xmath190 .
recall that any closed orientable surface with a negative euler number admits a metric of constant negative curvature @xmath191 .
so if @xmath189 , then @xmath192 admits such hyperbolic metric .
let @xmath193 denote the hyperbolic volume of @xmath192 , and let @xmath194 denote the volume of an ideal hyperbolic triangle @xmath195 in the hyperbolic plane @xmath196 . in @xmath0 ,
a remarkable convergence of topology and geometry takes place . in the spirit of this convergence , since @xmath197 , corollary 4.1 admits a more geometric reformulation : [ th1.3 ] let @xmath39 be a boundary generic vector field on a compact connected and orientable surface @xmath35 with boundary .
assume that @xmath189 .
then the complexity of the @xmath13-flow satisfies the inequality : @xmath198 moreover , @xmath199 if and only if @xmath13 is boundary concave . @xmath70 theorem [ th1.3 ] admits far reaching multidimensional generalizations ( see @xcite , @xcite ) .
they are valid for so called _ traversally generic _ vector fields ( see definitions 5.1 and 5.2 and @xcite ) on arbitrary smooth compact @xmath166-dimensional manifolds @xmath35 with boundary . such fields @xmath13 naturally generate stratifications of trajectory spaces @xmath34 , whose strata are labeled by the combinatorial patterns of tangency from the universal partially ordered set @xmath200}$ ] ( see the end of section 6 and @xcite ) . in high dimensions
, we use the simplicial semi - norms @xmath201 of gromov @xcite on the homology @xmath202 and @xmath203 ( as a substitute of the hyperbolic volume ) to provide lower bounds on the number of connected components of the @xmath200}$]-strata of any given dimension .
we say that a vector field @xmath39 on a compact surface @xmath35 is _ traversing _ if all its trajectories are closed segments or singletons . ] .
@xmath70 each trajectory @xmath124 of a traversing field @xmath13 must reach the boundary both in positive and negative times : otherwise @xmath124 is not homeomorphic to a closed interval .
we denote by @xmath204 the space ( in the @xmath74-topology ) of all traversing fields on @xmath35 .
we denote by @xmath205 the space ( in the @xmath74-topology ) of all gradient - like fields on a given compact surface @xmath35 and by @xmath206 the space of all non - vanishing fields on @xmath35 .
the next lemma says that @xmath13 is traversal if and only if it is non - vanishing and of a gradient type ( see @xcite for the proof ) . [ 5.1 ] for any compact connected surface @xmath35 with boundary , @xmath207 @xmath70 the surfaces @xmath35 and vector fields @xmath13 we consider are all smooth .
we can add an external collar to @xmath35 to form a diffeomorphic surface @xmath208 and to extend @xmath13 to a smooth field @xmath110 on @xmath209 .
let @xmath210 be a @xmath110-trajectory ( or rather its germ ) through a point @xmath10 of @xmath36 .
we can talk about _ order of tangency _ of two smooth curves , @xmath211 and @xmath36 , at @xmath212 in @xmath209 ( see definition [ def1.5 ] ) .
we say that the tangency of @xmath211 to @xmath36 is _ simple _ if its degree is 2
. when the two curves are transversal at @xmath10 we say that the order of tangency is @xmath50 .
in fact , this notions depend only on @xmath213 and not on the extension @xmath214 .
[ def1.4 ] a traversing vector field @xmath13 on a compact surface @xmath35 is called _ traversally generic _ ,
if two properties are valid : * ( 1 ) * if a trajectory @xmath124 is tangent to the boundary @xmath36 , then the tangency is simple , and * ( 2 ) * no @xmath13-trajectory @xmath124 contains more then one simple point of tangency to @xmath36 .
is boundary generic . ]
@xmath70 we denote by @xmath215 the space of all traversally generic vector fields on a compact surface @xmath35 .
in fact , the notion of traversally generic field is available in any dimension ( see @xcite ) .
as the name suggests , the traversally generic fields are typical among all traversing fields ; furthermore , a perturbation of any traversally generic field is traversally generic .
this is the content of the next theorem .
its validation requires an involved argument , which even in @xmath0 resists a significant simplification @xcite .
[ th1.4 ] for any compact connected surface @xmath35 with boundary , the space @xmath215 traversally generic fields is open and dense in the space @xmath216 .
we start with a couple of very natural questions . * question 6.1 .
* which compact connected surfaces with boundary admit boundary concave gradient - like vector fields @xmath39 ?
@xmath70 recall that @xmath217 .
* question 6.2 . *
are there compact connected surfaces @xmath35 with boundary for which @xmath218 ?
@xmath70 on many occasions we took advantage of the fact that , for traversally generic vector fields @xmath13 , the trajectory spaces @xmath34 are finite graph whose verticies have valency @xmath50 and @xmath51 only ( see fig .
4 ) . moreover , for a traversally generic boundary concave field @xmath13 , all the verticies of @xmath34 have valency @xmath51 .
now we will take a closer look at the graph - theoretical models of the boundary concave and traversally generic fields in 2d .
let @xmath219 be a finite connected trivalent graph with @xmath17 verticies .
we denote by @xmath220 its barycentric subdivision : each edge @xmath221 of @xmath219 is divided by a new vertex @xmath222 , its center .
we consider the finite set @xmath223 of all _ colorings _ of the edges of @xmath220 with _ tree _
colors so that , at each vertex of @xmath219 , exactly three distinct colors are applied .
thus , @xmath224 .
[ th1.5 ] let @xmath219 be a finite connected trivalent graph .
each coloring @xmath225 produces ( in a canonical way ) a compact connected surface @xmath226 with boundary .
the surface @xmath226 admits a traversally generic _ concave _ vector field @xmath227 .
the cardinality of the locus @xmath228 is the number of verticies in @xmath219 .
moreover , every connected surface with boundary , which admits a traversally generic concave vector field , can be produced in this way .
[ fig1.6 ] let @xmath229 denote the three distinct colors , and @xmath230 the entire pallet .
consider a @xmath169-dimensional space @xmath231 .
it has singularities in the form of binders of the - page open books ( see fig.1.6 ) .
the binders correspond to the verticies of @xmath219 .
first , employing a given coloring @xmath22 , we will construct a piecewise linear surface @xmath232 . the vector field on @xmath233
will be induced by the product structure in @xmath76 . for each edge @xmath234 and its barycenter @xmath235
, we place the interval @xmath236\subset z$ ] over @xmath222 .
let @xmath237 be half of the interval @xmath234 , bounded two vericies @xmath235 and @xmath238 . over @xmath237 ,
we place a strip @xmath239 ; its construction depends on the color attached to the interval @xmath240 $ ] as follows : * if the color of @xmath241 $ ] is @xmath242 , then we link the vertex @xmath243 with the vertex @xmath244 by a line in the rectangle @xmath245 \times ( 0 , 4)$ ] , and the vertex @xmath246 with the vertex @xmath247 by another line in @xmath248 ; * if the color of @xmath240 $ ] is @xmath249 , then we link by a line in @xmath248 the vertex @xmath250 with the vertex @xmath247 , and the vertex @xmath246 with the vertex @xmath251 by another line ; * if the color of @xmath240 $ ] is @xmath252 , then we link by a line in @xmath248 the vertex @xmath243 with the vertex @xmath244 , and the vertex @xmath246 with the vertex @xmath253 by another line . by definition , @xmath254 is the strip in @xmath240 \times ( 0 , 4)$ ] , bounded by the two lines whose construction is has been described above .
thanks to the monotonicity of the bijections @xmath255 , @xmath256 , and @xmath257 that correspond to the colors @xmath229 , the lines that bound the strips @xmath254 do not intersect .
we denote by @xmath233 the union of all such strips .
the local model of each binder implies that indeed @xmath233 is piecewise linear surface , imbedded in the singular space @xmath76 .
inside @xmath76 , one can smoothen the sharp edges of the boundary @xmath258 in order to get a smooth surface @xmath226 ( can you visualize this smoothing in the vicinity of point @xmath247 from fig . 6 ? ) .
the restriction of the product structure in @xmath76 to its subspace @xmath226 produces a smooth non - vanishing vector field @xmath227 on @xmath226 .
its trajectories ( the vertical lines in @xmath76 ) will be simply tangent to @xmath259 exactly at the points of the type @xmath260 , where @xmath261 runs over the set of verticies of @xmath219 . by theorem [ th1.3 ] , this field @xmath227 is of the gradient type .
conversely , any traversally generic and concave vector field @xmath13 on a connected compact surface @xmath35 with boundary , produces a map @xmath52 , where the space of trajectories is a finite trivalent graph .
its verticies are in 1-to-1 correspondence with the points of the locus @xmath125 . as a point @xmath222 in the open edge @xmath221 of the graph
@xmath34 approaches a vertex @xmath261 , the intersection of the @xmath13-trajectory @xmath262 with the boundary @xmath36 defines a bijection of the @xmath13-ordered set @xmath263 of cardinality @xmath169 to a @xmath13-ordered set @xmath264 of cardinality @xmath51 , the orders being respected by the bijections .
this determines one of three colors we attach to the half - edge @xmath240 $ ] .
therefore the geometry of the flow determines a tricoloring of the graph @xmath265 .
the next theorem answers questions 6.1 and 6.2 .
[ th6.2 ] * for any orientable connected surface @xmath35 with boundary , the two complexities are equal : @xmath266 .
moreover , any such @xmath35 , but the disk , admits a boundary concave traversally generic vector field . as a result , for any orientable connected @xmath35 with boundary , but the disk , the two complexities are equal to @xmath267 . * for any
non - orientable connected surface @xmath35 with boundary , which is a boundary connected sum of several punctured klein bottles and annuli , @xmath268 as well .
again , any such @xmath35 admits a boundary concave traversally generic vector field . * in contrast , the mbius band @xmath176 does not admit a boundary concave traversally generic vector field .
in fact , @xmath178 and @xmath177 .
* moreover , @xmath269 and @xmath270 for any @xmath35 as in the first two bullets .
consider the boundary connected sum @xmath271 of two compact surfaces @xmath272 and @xmath273 . the euler number of the sum satisfies the rule @xmath274 on the other hand , given two boundary generic fields @xmath275 and @xmath276 , there exists a traversally generic field @xmath261 on @xmath271 such that @xmath277 indeed , we may attach a @xmath50-handle @xmath278 to @xmath279 so that an @xmath278 has a neck with respect to the extension @xmath261 .
such field @xmath261 contributes two points to @xmath280 .
of course , this construction fails when @xmath281 ; however , for traversing fields @xmath13 , both loci @xmath282 . by corollary 4.1 ,
if @xmath35 admits a boundary concave field , then @xmath283 , provided @xmath180 .
in particular , if @xmath35 with a non - positive euler number admits a _ boundary concave _ traversally generic @xmath13 , then @xmath284 let @xmath275 and @xmath276 be some boundary generic/ traversally generic fields which deliver the two gradient complexities .
the previous arguments about extending @xmath275 and @xmath276 across the handle @xmath278 imply that if @xmath285 and @xmath286 ( say both surfaces admit boundary concave and traversally generic fields ) , then @xmath287 provided that @xmath288 .
since the reverse inequality holds by corollary 4.1 , we get @xmath289 when @xmath285 and @xmath286 .
recall the topological classification of closed connected surfaces .
any such surface is either a sphere , or a connected sum of several tori ( the orientable case ) , or a connected sum of several projective spaces ( the non - orientable case ) .
therefore any connected surface with boundary is obtained from the surfaces in this list by deleting at least one disk .
let @xmath290 denote the complement to an open disk in a @xmath169-torus , and @xmath176 denote the complement to an open disk in a projective plane the mbius band , and let @xmath291 denote the annulus .
thus any connected surface with boundary is either a disk @xmath292 , or a boundary connected sum of several copies of punctured tori @xmath290 and annuli @xmath291 ( the orientable case ) , or a boundary connected sum of several copies of mbius bands @xmath176 and annuli @xmath291 ( the non - orientable case ) .
let us now compute the complexities of the basic blocks in this decomposition .
note that @xmath293 , since @xmath292 admits a convex traversing flow .
also @xmath294 , the latter equality being delivered by the radial gradient field .
we claim that @xmath295 . indeed , since @xmath296 , by corollary 4.1
, we get @xmath297 . on the other hand ,
there exists a trivalent graph @xmath298 with an appropriate tricoloring and exactly _ two _ verticies such that , applying the construction from theorem [ th1.5 ] , we produce a traversally generic field @xmath299 on the surface @xmath300 with the cardinality @xmath169 locus @xmath301 . as a result , both complexities of @xmath290 equal to @xmath169 .
similar considerations apply to the punctured klein bottle @xmath302 and a different trivalent graph @xmath303 with two verticies and an appropriate tricoloring . since @xmath304
, we conclude that @xmath305 . the third trivalent graph @xmath306 with two verticies and an appropriate tricoloring delivers a traversally generic boundary concave flow on a punctured annulus @xmath307 , the disk with two holes .
thus , @xmath308 .
in fact , theorem [ th1.5 ] implies that @xmath309 are the only connected surfaces of the gradient complexity @xmath169 that admit concave traversally generic fields .
indeed , just start with the tree @xmath310 " with two trivalent verticies and consider the ways one can identify its four leaves in pairs . then consider all admissible tricologings of the resulting graphs @xmath219 .
this cases will deliver the three model tricolored graphs @xmath311 .
now the quasi - additivity " of euler numbers and gradient complexities under the connected sum operations imply that the gradient complexity of boundary connected sums @xmath312 of several copies of the model surfaces @xmath313 is equal to @xmath314 . indeed
, these properties imply that @xmath315 , while in general @xmath316 .
moreover , every such surface @xmath35 admits a _
boundary concave _ traversally generic field ( by the @xmath50-handle - with - a - neck argument ) , since the basic blocks @xmath317 and @xmath291 do .
the mbius band @xmath176 is different .
we notice that @xmath176 admits a non - vanishing vector field @xmath13 with a single closed trajectory the core of the mbius band and transversal to the boundary @xmath318 .
thus , @xmath178 . now consider a trivalent graph @xmath319 with a single vertex of valency @xmath51 and a single vertex of valency @xmath50 ( this @xmath319 is a circle to which a radius is attached ) .
the construction from theorem [ th1.5 ] applies to produce a remarkable embedding of the mbius band in the product @xmath320 $ ] .
so we conclude that @xmath176 admits a traversally generic field @xmath13 ( not concave ! ) with @xmath125 being a singleton ( @xmath321 is a singleton as well ) . as a result , @xmath322 . on the other hand ,
any traversally generic field @xmath13 on @xmath176 must produce the graph @xmath34 which is homotopy equivalent to a circle , the homotopy type of @xmath176 .
if @xmath323 , this graph @xmath34 has no trivalent verticies , in which case , @xmath34 is homeomorphic to a circle .
so @xmath324 must be a fibration whose fibers ( the @xmath13-trajectories ) are segments .
moreover , thanks to the field @xmath13 , this fibration is orientable , a contradiction with the non - orientability of @xmath176 . therefore , we conclude that @xmath177 , while @xmath178 . finally , for any @xmath35 which is a boundary connected sum of @xmath290 s , @xmath325 s , and @xmath291 s , by the same arguments , the inequalities @xmath326 and @xmath327 hold .
this validates the claim in the last bullet .
pick an extension @xmath209 of a given compact surface @xmath35 by adding an external collar to @xmath35 .
let @xmath110 be an extension of a given field @xmath13 into @xmath209 .
pick a smooth auxiliary function @xmath328 such that : * @xmath329 is a regular value of @xmath8 , * @xmath330 , * @xmath331 ) = x$ ] , @xmath332 [ def1.5 ] let @xmath211 be a @xmath110-trajectory through a point @xmath333 .
we say that @xmath211 has the _ order / multiplicity of tangency _
@xmath95 to @xmath36 at @xmath10 , if @xmath334 for all @xmath335 , and @xmath336 at @xmath10.-st jet at @xmath10 of @xmath337 vanishes , but the @xmath95-th jet does not . ] here @xmath338 denotes the @xmath339 iterated @xmath110-directional derivative of the function @xmath8 .
@xmath70 given a traversally generic vector field @xmath13 on a compact connected surface @xmath35 , we will attach the _
combinatorial pattern _
@xmath340 to a typical @xmath13-trajectory @xmath341 that corresponds to the edges of the graph @xmath34 , the pattern @xmath342 to the trajectories that correspond to the trivalent verticies of @xmath34 , and the pattern @xmath343 to the univalent verticies ( see fig .
4 ) . in fact , the numbers 1 and 2 in these patterns reflect the order of tangency of the curves @xmath211 and @xmath36 at the points of @xmath263 ( see definition [ def1.5 ] ) . on a given compact surface @xmath35 , for _
traversally generic _ fields @xmath13 no other patterns ( say , like @xmath344 or @xmath345 ) occur . in @xmath0 ,
this conclusion follows from definition [ def1.5 ] .
the lemma below is another way to state this fact .
its proof , relying on the malgrange preparation theorem @xcite , can be found in @xcite .
let @xmath13 be a traversally generic field on @xmath35 .
extend @xmath213 to a pair @xmath214 . in the vicinity of each @xmath13-trajectory @xmath124
, there exist special local coordinates @xmath346 in @xmath209 and a real polynomial @xmath347 of degree @xmath169 or @xmath48 such that : * each @xmath110-trajectory is given by the equation @xmath348 , * the boundary @xmath36 is given by the polynomial equation @xmath349 , * @xmath35 is given by the polynomial inequality @xmath350 .
the polynomial @xmath347 takes three canonical forms : 1 .
@xmath351 , which corresponds to the combinatorial pattern @xmath352 , 2 .
@xmath353 which corresponds to the combinatorial pattern @xmath354 , 3 .
@xmath355 which corresponds to the pattern @xmath356 .
@xmath70 to summarize , at @xmath357 the order of tangency is @xmath169 ; the trajectories through @xmath125 have the combinatorial tangency pattern @xmath342 , and through @xmath321 the combinatorial tangency pattern @xmath343 .
the rest of trajectories have the pattern @xmath358 .
we denote by @xmath359}$ ] the partially ordered set whose elements are @xmath360 and the order is defined by @xmath361 and @xmath362 .
this combinatorics does not look impressive .
however , in higher dimensions , traversally generic fields on @xmath166-manifolds with boundary generate a rich and interesting partially ordered finite list @xmath200}$ ] of combinatorial tangency patters .
the poset @xmath200}$ ] is _ universal _ in each dimension @xmath363 .
they are discussed in @xcite .
let @xmath13 be a traversing and boundary generic vector field on a compact connected surface @xmath35 with boundary .
for any point @xmath364 , consider the closest point @xmath365 that can be reached by moving along the trajectory @xmath366 through @xmath8 in the direction of @xmath13 ( see fig .
note that @xmath367 if and only if @xmath368 .
the correspondence @xmath369 defines a map @xmath370 which we call the _ causality map_. it is a distant relative of the classical poincar return map .
alternatively , one can think of @xmath371 as determining a _ partial order _
@xmath372 " among the points of the boundary @xmath36 .
the word causality " in the name of @xmath371 is motivated by the following pivotal special case .
[ fig1.7 ] in the vicinity of @xmath10 . , width=384,height=259 ] * example 8.1 . *
let @xmath373 be a smooth time - dependent vector field on the circle @xmath374 ( equipped with the angular coordinate @xmath375 ) .
it gives rise to a vector field @xmath376 on the cylinder @xmath377 .
we think about the factor @xmath374 as _ space _ and about the factor @xmath378 as _ time _
so we call @xmath377 _ the space of events_. note that @xmath39 is a gradient - like field with respect to the time function @xmath379
. pick any smooth compact and connected surface @xmath380 .
such a surface has a boundary @xmath36 .
we call @xmath35 the _ event domain _ , and its boundary @xmath36 the _ event horizon_. since the field @xmath39 is traversing in @xmath35 , the map @xmath371 is well - defined
. then the map @xmath381 indeed gives rise the _ causality relation _ on the event horizon : the correspondence @xmath371 reflects the _ evolution _ of an event @xmath8 into the event @xmath382 .
@xmath70 let @xmath383 denotes the union of @xmath13-trajectories through the points of the concavity locus @xmath125 .
the causality map is _ discontinuous _ at the points of the intersection @xmath384 ( see fig .
7 ) . on the positive side , the discontinuities of the causality map @xmath371 are not too bad : in a sense , the map has left " and right " limits . given a pair @xmath213 , the @xmath13-trajectories , viewed as _ _
un__parametrized @xmath13-oriented curves , produce an oriented @xmath50-dimensional foliation @xmath385 on @xmath35 . [ th7.1]*(the causal holography principle in 2d ) . * let @xmath386 and @xmath387 be two compact connected surfaces with boundaries , carrying traversally generic vector fields @xmath275 and @xmath276 , respectively .
assume that there is a diffeomorphism @xmath388 which conjugates the two causality maps : @xmath389 then @xmath390 extends to a diffeomorphism @xmath391 which maps the oriented foliation @xmath392 to the oriented foliation @xmath393 .
we will only sketch the argument . a fully developed proof of the multidimensional analogue of this theorem is contained in @xcite .
first , we notice that since @xmath394 and @xmath395 are @xmath390-conjugate , the diffeomorphism @xmath390 induces a well - defined continuous map @xmath396 of the trajectory spaces .
moreover , @xmath397 preserves the stratifications of the two trajectory spaces / graphs by the combinatorial type of trajectories .
that is , the trivalent verticies of @xmath398 are mapped to the trivalent verticies of @xmath399 , the univalent verticies are mapped to univalent verticies , and the interior of the edges to the interior of the edges .
then we pick a smooth function @xmath400 such that @xmath401 . with the help of @xmath390 , we pull - back @xmath402 to get a smooth function @xmath403 such that @xmath404 for all @xmath405 .
then we argue that @xmath406 extends to a smooth function @xmath407 such that @xmath408 .
we use @xmath407 to embed @xmath272 in the product @xmath409 by the formula @xmath410 where @xmath366 , the @xmath275-trajectory through @xmath8 , is viewed as the point @xmath411 of the graph @xmath398 . similarly , we use @xmath412 to embed the surface @xmath273 in the product @xmath413 with the help of the map @xmath414 .
finally , we employ @xmath397 , @xmath407 and @xmath412 to construct a map @xmath415 by the formula @xmath416 where @xmath33 belongs to the @xmath412-image of the trajectory @xmath417 . crudely ,
the restriction of @xmath418 to @xmath419 is the desired diffeomorphism @xmath391 . note that , in general , the pull - back @xmath420 is not @xmath407 ; so the parametrizations of the trajectories are not respected by the diffeomorphism @xmath421 , but the @xmath50-foliations @xmath392 and @xmath393 are . [ cor7.1 ] let @xmath35 be a compact connected surface with boundary , and @xmath13 a smooth traversally generic vector field on it .
then the knowledge of the causality map @xmath381 is sufficient for a reconstruction of the pair @xmath422 , up to a diffeomorphism that is constant on @xmath36 . @xmath70
the world holography " is present in the name of theorem [ th7.1 ] since the surface @xmath35 and the @xmath0-dynamics of the @xmath13-flow in it are recorded on two @xmath50-dimensional screens , @xmath57 and @xmath58 .
theorem [ th7.1 ] and corollary [ cor7.1 ] are valid in any dimension ( @xcite ) .
* example 8.2 .
* let @xmath13 be a traversally generic field on a connected surface @xmath35 whose boundary @xmath36 is a _ single _ loop
. then the boundary @xmath36 is divided into @xmath423 disjoint arcs @xmath424 that form @xmath57 and @xmath423 complementary arcs @xmath425 that form @xmath58 .
the causality map @xmath426 can be represented by its graph @xmath427 . the map @xmath371 ( the curve @xmath428 ) is discontinuous at exactly @xmath429 points in @xmath430 that correspond to the points of the intersection @xmath384 . there
the map @xmath371 has distinct left and right limits . according to the corollary [ cor7.1 ]
, the curve @xmath427 determines @xmath35 and the _ un_-parametrized dynamic of the @xmath13-flow , up to a diffeomorphism @xmath431 that is the identity on @xmath36 .
note that the number @xmath423 alone is not sufficient even to determine the genus of the surface @xmath35 .
@xmath70 revisiting example 8.2 , we get the following interpretation of corollary [ cor7.1 ] : [ cor7.2 ] for any smooth time - dependent vector field @xmath261 on the circle @xmath374 , the causality relation on the event horizon @xmath36 is sufficient for a reconstruction of the event domain @xmath35 and the un - parametrized dynamics of the @xmath432-flow , up to a diffeomorphism of @xmath35 that is the identity on @xmath36 . @xmath70
the theory of billiards on riemmanian surfaces @xmath35 with boundary benefits from applying the @xmath433-version of the causal holography to the geodesic flow on the @xmath51-fold @xmath434 , the space of unit tangent vectors on @xmath35 .
see @xcite for some of these applications .
in addition to geodesic billiards , they include the classic inverse geodesic scattering problems .
traversing flows have interesting _ characteristic classes_elements of certain cohomology associated with them . in dimension two
, they are quite primitive , but for high - dimensional flows , surprisingly rich ( see @xcite ) .
we have seen that the traversally generic flows exhibit a very particular combinatorial patterns of tangency to the boundary @xmath36 . in particular , for generic @xmath0-flows , no tangencies of orders @xmath435 occur .
there is a nice link between this behavior and the _ spaces of smooth functions _ @xmath436 or even polynomials that have no zeros of multiplicities @xmath435 . to explain the connection
, we will need the following definition / construction .
let @xmath437 denote the space ( in the @xmath74-topology ) of smooths functions @xmath436 which are identically @xmath50 outside of a compact set .
let @xmath438 be its subspace , formed by functions that have zeros only of multiplicity @xmath439 .
such spaces of functions with _ moderate singularities _ " have been studied in depth by v. i. arnold @xcite and v. a. vassiliev @xcite . in @xmath0 , we employ just a tiny portion of their results . the main theorem of arnold - vassiliev describes the weak homotopy / homology types of the spaces @xmath440 for all @xmath441 . in particular , the homology of the space @xmath438 is isomorphic to the homology of @xmath442 , the space of loops on a @xmath169-sphere ( @xcite ) !
arnold proved also that the fundamental group @xmath443 @xcite .
for an even non - negative integer @xmath6 , we will also explore the subspaces @xmath444 , formed by functions whose _ degree_the sum of multiplicities of all its zeros is even and does not exceed @xmath6 .
let @xmath110 be a boundary convex traversing vector field on an annulus @xmath291 . with the help of @xmath110
, we can introduce a product structure @xmath445 $ ] so that the fibers of the projection @xmath446 are the @xmath110-trajectories .
we say that a boundary convex traversing vector field @xmath110 is _ generic with respect to _ @xmath447 , if no @xmath110-trajectory @xmath124 contains more than one point of self - intersection from @xmath447 and no more than one point of simple tangency to @xmath447 , but not both .
@xmath70 for a given @xmath447 , by standard techniques of the singularity theory , we can find a perturbation of @xmath110 within the space @xmath448 so that the perturbed field is generic with respect to @xmath447 . since an _ immersion _ is a smooth map of manifolds , whose differential has the trivial kernel , the immersions allow for a _ transfer _ of a given vector field on the target manifold to a vector field in the source manifold .
the transfer of a non - vanishing field is a non - vanishing field .
all surfaces in this section are orientable .
note that any orientable surface @xmath35 admits an _ immersion _ @xmath449 ( or even in the plane @xmath49 ) ( see fig .
we will use this fact to pull - back non - vanishing fields on the target space @xmath291 to @xmath35 .
[ def1.7 ] consider an _ immersion _
@xmath449 of a given compact orientable surface @xmath35 into an annulus @xmath291 , equipped with a traversal boundary convex ( radial " ) field @xmath110 .
we call such @xmath22 _ generic relative to _ @xmath110 , if @xmath110 is generic with respect to the curves @xmath133 in the sense of definition 9.1 . given a transversally generic field @xmath13 on a connected compact surface @xmath35 , we call a map @xmath450 a _ convex quasi - envelop of _
@xmath213 if there exists an immersion @xmath449 which is generic relative to the radial field @xmath110 on @xmath291 , and @xmath146 , the pull - back of @xmath110 .
@xmath70 given a boundary generic relative to @xmath110 immersion @xmath449 , the @xmath22-pullback ( transfer ) of the field @xmath110 defines a vector field @xmath39 on @xmath35 . since @xmath22 is an immersion , evidently the pull - back @xmath13 is traversing on @xmath35 .
moreover , @xmath13 is taversally generic in the sense of definition [ def1.4 ] , since no @xmath13-trajectory @xmath124 has more than one point of simple tangency to @xmath36 .
[ def1.8 ] let @xmath449 be a
_ regular embedding _ of a given compact surface @xmath35 into an annulus @xmath291 , carrying a traversal boundary convex field @xmath110 .
we denote by @xmath13 the pull - back of @xmath110 under @xmath22 .
if @xmath22 is traversally generic relative to @xmath110 , then we say that the pair @xmath451 is a _ convex envelop _ of @xmath213 .
if a compact connected surface @xmath35 with boundary has a pair of loops whose transversal intersection is a singleton , then no traversal flow on @xmath35 admits a convex envelop . in other words , if a connected surface @xmath35 with boundary has a handle , then no traversal flow on @xmath35 can be convexly enveloped . by lemma 1.2 ,
the space @xmath209 of a convex envelop is either a disk or an annulus , both surfaces residing in the plane .
no two loops in the plane intersect transversally at a singleton .
thus , for surfaces with a handle , no convex envelops exist .
so the existence of a convex envelop severally restricts the topology of surface @xmath35 . to incorporate surfaces with handles into our constructions ,
we have introduced the notion of a convex quasi - envelop ( definition [ def1.7 ] ) .
now we are in position to explore a connection between immersions @xmath452 of a given surface @xmath35 in the annulus @xmath291 , such that @xmath146 and @xmath110 is generic with respect to @xmath133 on one hand , and loops in the functional spaces @xmath438 on the other .
* @xmath457 , * @xmath329 is the regular value of @xmath458 at the points of @xmath454 , * in the vicinity of each point @xmath459 , consider local coordinates @xmath460 such that @xmath461 and @xmath462 define the two intersecting branches of @xmath133 ; then locally @xmath463 , where the constant @xmath464 .
* @xmath465 in the vicinity of @xmath466 , * the sign of @xmath458 changes to opposite as a path crosses an arc from @xmath454 transversally provides a checker board " coloring of the domains in @xmath467 . ] .
here we denote by @xmath307 the interior of the annulus @xmath291 .
let @xmath469 be a smooth function so that @xmath470 in @xmath307 and @xmath471 for all @xmath110trajectories @xmath211 in @xmath291 .
then , with the help of @xmath458 and @xmath472 , we get a map @xmath473 whose target is the space of smooth functions @xmath436 with no zeros of multiplicity @xmath474 and that are identically @xmath50 outside of a compact set in @xmath378 .
we define the map @xmath475 by the formula @xmath476 where , abusing notations , @xmath210 stands for both a @xmath110-trajectory in @xmath291 and for the corresponding point in the trajectory space @xmath477 . for a fixed @xmath22 , it is easy to check that the homotopy class @xmath478 $ ] of @xmath475 does not depend on the choice of the auxiliary function @xmath458 , subject to the five properties in ( [ eq1.7 ] ) ( the space of such @xmath458 s is convex and thus contractible ) .
we pick a generator @xmath479 ( see @xcite ) and define the integer @xmath480 by the formula @xmath481 $ ] . as a result ,
any immersion @xmath449 , which is generic with respect to @xmath110 , produces a homotopy class @xmath478 \in \pi_1(\mathcal f_{\leq 2})$ ] and an integer @xmath480 .
the isomorphism @xmath443 follows from the work of v. i. arnold @xcite by a slight modification of his arguments , which we will describe next ( see theorem [ th1.7 ] ) .
the main difference between our constructions and the ones from @xcite is that arnold uses the critical loci of functions from @xmath438 , while we are using the zero loci .
generic loops in @xmath482 have an interpretation in terms of finite collections @xmath483 of smooth closed curves in the annulus @xmath291 with _ no inflection points with respect to their tangent lines of the form _
@xmath484 in the @xmath485-coordinates .
we call such tangent lines @xmath375-_vertical_. furthermore , the generic homotopy between such loops @xmath486 correspond to some _ cobordism _ relation between the corresponding plane curves , the cobordism also avoids the @xmath375-vertical inflections . 1 .
@xmath487 is a finite collectionof closed smooth _
curves @xmath488 , 2 . the projections @xmath489
have morse type singularities only - vertical inflections . ] , 3 .
the self - intersections and mutual intersections of the curves @xmath488 are transversal and no triple intersections are permited , 4 . at each double intersection ,
the two banches of @xmath483 are not parallel to the @xmath490-coordinate , 5 .
the @xmath375-images of the intersections and of the critical values of @xmath489 are all distinct in @xmath374 , 6
. the cardinality of each fiber of @xmath491 does not exceed a given natural number @xmath6 .
[ def1.9 ] given two collections @xmath493 and @xmath494 of immersed closed curves as in ( [ eq1.9 ] ) , we say that they are _ cobordant with no @xmath375-vertical inflections _ , if there is a smooth function @xmath495 \to \r$ ] such that : * @xmath329 is a regular value of @xmath496 , * the restriction of the projection @xmath497 \to [ 0 , 1]$ ] to the zero set @xmath498 is a morse function , * @xmath499 and @xmath500 , * for each @xmath501 $ ] , the section @xmath502 is such that @xmath503 has no @xmath375-horizontal inflections . ] * for each @xmath501 $ ] the cardinality of the fibers of @xmath504 does not exceed a given natural number @xmath6 .
@xmath70 it is possible to verify that the cobordism with no @xmath375-vertical inflections is an _ equivalence relation _ among collections of curves as in ( [ eq1.9 ] ) . indeed ,
if @xmath483 is cobordant to @xmath505 with the help of @xmath496 , and @xmath505 to @xmath506 with the help of @xmath507 , then there exists a piecewise smooth function @xmath508 \to \r$ ] whose restriction to @xmath509 $ ] is @xmath496 and to @xmath510 $ ] is a @xmath511-shift of @xmath507 . smoothing @xmath512 along @xmath513 in the normal direction and scaling down the interval @xmath514 $ ] to @xmath515 $ ] , produces the desired function - cobordism @xmath516 \to \r$ ] .
so we can talk about _ the set of bordisms _
@xmath517 , based on collections of closed curves in the annulus with no @xmath375-vertical inflections .
this set is a _ group _ : the operation @xmath518 is defined by the union @xmath519 , where @xmath520 and @xmath521 are the images of @xmath483 and @xmath505 , scaled down in the @xmath490-direction by the factor @xmath522 and placed in sub - annuli of @xmath523 $ ] .
the role of @xmath524 is played by the mirror image of @xmath483 with respect to a vertical ( equivalently , horizontal ) line , a fiber of @xmath525 .
note that this operation @xmath526 may affect the maximal cardinalities @xmath6 and @xmath527 of the fibers @xmath491 and @xmath528 in a somewhat unpredictable way . in any case , the fiber cardinality of @xmath529 has the upper boundary @xmath530 .
[ th1.7 ] the fundamental group @xmath531 is isomorphic to the bordism group @xmath517 , based on finite collections of immersed loops with no @xmath375-vertical inflections in the annulus @xmath291 and subject to the constraints ( [ eq1.9 ] ) .
the isomorphism is induced by the correspondence @xmath532 } \subset a.\ ] ] @xmath70
this theorem is a foundation of a _ graphic calculus _ that converts homotopies of loops in the functional space @xmath438 into cobordisms of closed loop patterns in the annulus @xmath291 with no @xmath375-vertical inflections . figures 10 - 14 show an application of this calculus .
they explain why any loop in @xmath438 is homotopic to an integral multiple of a generator @xmath533 , represented by a model loop pattern @xmath534 as in fig .
9 , diagram ( a ) or ( b ) .
we fix an orientation of @xmath35 , thus picking orientations for each component of @xmath36 .
given an orientation - preserving immersion @xmath452 such that @xmath133 has the properties as in ( [ eq1.9 ] ) , we notice that the polarity of @xmath154 is @xmath155 if and only if @xmath536 , where @xmath158 is the inner normal to @xmath36 at @xmath17 , points in the direction of @xmath375 . otherwise , the polarity of @xmath17 is @xmath159 ( see fig .
[ th1.8 ] any orientation - preserving immersion @xmath452 such that @xmath110 is generic with respect to @xmath133 of @xmath213 ] produces a map @xmath538 ( see ( [ eq1.8 ] ) ) .
its homotopy class @xmath478 = j^\a \cdot \kappa$ ] , where @xmath539 denots a generator of @xmath540 . for any @xmath110-generic immersion @xmath545
, we pick an auxiliary function @xmath546 , adjusted to @xmath22 as in ( [ eq1.9 ] ) . by the previous arguments , this choice produces the loop @xmath547 .
although the loop @xmath475 is generated by an immersion @xmath449 , in the process of deforming @xmath475 by a cobordism @xmath495 \to \r$ ] with no @xmath375-vertical inflections as in definition 9.4 , we may destroy this connection with the original @xmath22 : the new curve patterns @xmath548}$ ] in @xmath291 may not be produced by immersions @xmath549}$ ] .
let us describe an algorithm ( see figures 10 - 14 ) that reduces a given pattern @xmath550 to a pattern from _ the canonical set of patterns _ @xmath551 ( as in fig .
9 ) by a cobordism @xmath495 \to \r$ ] . we will perform a sequence of elementary surgeries on the set @xmath493 , executed inside of the cylindrical shell @xmath509 $ ] .
it is sufficient to construct a smooth surface @xmath552 $ ] as in definition 9.4 , for which @xmath553 and @xmath554 ; then one can define a function @xmath495 \to \r$ ] , appropriately adjusted to @xmath555 , so that @xmath329 is a regular value of @xmath496 and @xmath556 .
as we modify the @xmath33-section @xmath557 , we keep track of the checker board polarities @xmath558 , attached to the regions of @xmath559 ; through the process , the polarity of the region adjacent to @xmath466 remains @xmath145 " .
let us denote by @xmath560 the region of the negative polarity that is bounded " by the curve pattern @xmath557 .
@xmath561 denotes the complementary set .
informally , the regions of polarity @xmath145 are the regions where the function @xmath496 from definition [ def1.9 ] is non - negative . with the help of this polarization @xmath562 of the annulus @xmath291 , the points @xmath563 , where @xmath110-flow is tangent to @xmath503 , acquire the polarization
@xmath145 " or @xmath144 " : if the germ of the trajectory @xmath564 is contained in @xmath560 , then the polarity of @xmath17 is defined to be
@xmath145 " , otherwise it is @xmath144 " .
moreover , if the inner normal @xmath158 to the region @xmath565 at @xmath17 has the same direction as the coordinate @xmath375 on @xmath291 , then the second polarity of @xmath17 is defined to be
@xmath155 " , otherwise it is @xmath159 " . as a result , we can talk about the four sets : @xmath566 , @xmath567 , @xmath568 , @xmath569 .
we simplify the notations for these loci as : @xmath570 * ( 1 ) * at any stage of this construction , we can resolve each crossing @xmath571 in a preferred way .
the two branches of the preferred resolution will be transversal to the @xmath375-fiber through @xmath17 . when the vector @xmath572 points inside @xmath560 , the resolution will add a 1-handle to @xmath561 , when the vector @xmath572 points inside @xmath561 , the resolution will add a 1-handle to @xmath560 . in any case
, the sets @xmath573 , @xmath574 are not affected . as a result of these resolutions ,
the new pattern @xmath575 is a disjoint union of _ simple _ smooth curves with no @xmath375-vertical inflections .
moreover , it shares with @xmath493 the same sets @xmath576 , @xmath577 s , which eventually will be eliminated . ]
( see fig .
10 ) . *
( 2 ) * next we will pick and fix one regular value @xmath578 of the map @xmath504 .
the intersection of @xmath579 consists of several intervals . by performing @xmath50-surgery of @xmath503 along each of these intervals
we get a new loop pattern @xmath580 which has an empty intersection with the @xmath375-fiber over the point @xmath581 .
moreover no @xmath375-vertical inflections were introduced in the process .
the original loci @xmath576 , @xmath577 are preserved ( while the loci @xmath582 , @xmath583 are changed ) .
therefore we may assume that @xmath584 is contained in a rectangle @xmath585 and @xmath503 shares the numbers of points from the loci @xmath576 , @xmath577 with the original @xmath493 ( see fig .
11 , diagram 3 ) . *
( 3 ) * consider the set @xmath586 of critical values of @xmath504 for the critical points from @xmath587 .
we pick a regular value @xmath588 in - between each pair of adjacent critical values @xmath589 .
then we apply @xmath50-surgery on @xmath561 as in ( 2 ) to empty the region @xmath560 in the vicinity of the fiber @xmath590 ( see fig .
11 , diagram 4 ) . *
( 6 ) * pairs of points @xmath594 and @xmath595 can be cancelled via a surgery on their regions @xmath596 ( as shown in fig .
13 , diagrams 7 and 8 , and fig .
14 , diagrams 9 and 10 ) .
this cancellation of pairs will be executed gradually and with some care .
any strip @xmath597 , bounded by the vertical lines @xmath598 and @xmath599 , contains a single region @xmath600 with @xmath601 . if the points of opposite second polarity ( @xmath153 ) exist , then there are two _ adjacent _ vertical strips @xmath602 such that @xmath603 and @xmath604 have opposite second polarities .
we attach to @xmath605 two @xmath50-handles to form an annulus @xmath606 as in figures 12 and 13 . to complete the cancellation of opposite pairs , we perform @xmath169-surgery on the inner circles of the annuli @xmath607 .
this converts the annuli into disks , residing in @xmath560 .
they can be eliminated by @xmath329-surgery along the outer circles of @xmath607 s ( see fig .
14 , diagram 10 ) .
it may happen that the model domain @xmath608 as in fig .
9 , ( b ) , and its mirror image " @xmath609 with respect to a vertical line @xmath610 are positioned so that their horns are pointing in opposite directions .
in such a case , they can be cancelled by a slightly different sequence of elementary surgeries ( see @xcite ) .
alternatively , taking a trip around the annulus @xmath291 , we will find a pair of adjacent strips such that their domains @xmath600 and @xmath611 of opposite second polarity can lock horns " . for them , the previous recipe will apply .
this cancellation procedure can be repeated by considering the remaining adjacent pairs of regions with the opposite second polarity untill no regions with the opposite second polarity are left . * ( 7 ) * as a result of all these steps , @xmath565 is either empty , or a disjoint union of disks ( as in fig .
9 ) , each of which contains a single point from @xmath593 ( and tree points from @xmath612 ) ; _ the second polarities of such points are the same for all disks_. thus we got an integral multiple of the basic pattern as in fig .
9 and proved that @xmath613 .
note that the original difference @xmath614 between the numbers of @xmath110-trajectories with polarities @xmath155 and @xmath159 and of the combinatorial types @xmath615 is preserved under the modifications in ( 1)-(7 ) .
* remark 9.1 .
* it is interesting and somewhat surprising to notice that the invariant @xmath618 reflects more the topology of the field @xmath619 than the topology of the surface @xmath35 : in fact , any integral value of @xmath480 can be realized by a traversally generic field @xmath13 on a _ disk _
@xmath292 which even admits a convex envelop !
a portion of the boundary @xmath620 looks like a snake with respect to the field @xmath110 of the envelop . for any @xmath35 , the effect of deforming a portion of @xmath36 into a snake is equivalent to adding several times a spike ( an edge and a pair of univalent and trivalent verticies ) to the graph @xmath34 .
evidently , these operations do not affect @xmath621 .
for example , for @xmath22 as in fig .
8 , @xmath623 .
if we subject @xmath22 to an isotopy that introduces a snake - like pattern of fig .
9 , ( a ) , then for the new immersion @xmath624 , the invariant @xmath625 . @xmath70 * remark 9.2 . *
consider a connected oriented surface @xmath35 with a connected boundary .
it is a boundary connected sum of a few copies of @xmath290 , the torus with a hole .
a punctured torus admits an immersion @xmath626 in the annulus so that the cardinality of the fibers of @xmath627 does not exceed @xmath628 ( see fig . 8) .
therefore , any connected oriented surface @xmath35 with boundary admits an immersion @xmath449 with the property @xmath629 for all @xmath578 . @xmath70
the finite - dimensional space @xmath630 of real monic polynomials of an even degree @xmath6 and with no real roots of multiplicity @xmath435 is a natural approximation " of the functional space @xmath631 .
of course , a polynomial from @xmath630 is not a function from @xmath438 : it is not identically @xmath50 outside of a compact set .
however , there is an embedding @xmath632 that , in the vicinity of @xmath633 , levels down to @xmath50 " any real polynomial @xmath634 of an even degree @xmath6 .
its image belongs to the subspace @xmath631 .
this embedding is described by an analytic formula ( see @xcite ) as follows .
fix an auxiliary smooth function @xmath635 $ ] such that @xmath636 for @xmath637 , @xmath638 for @xmath639 , and @xmath640 for @xmath641 .
let @xmath642 denote the sum of absolute values of the coefficients of the monic @xmath634
. then @xmath643 consider the forbidden set " @xmath645 of functions @xmath3 that have at least one zero of multiplicity @xmath435 . among them , the functions @xmath3 that have exactly one zero of multiplicity @xmath51 form an open and dense subset @xmath646 .
for any @xmath651 , consider a @xmath375-family of quartic @xmath490-polynomials @xmath652\big\}_{\theta \in [ 0 , 2\pi]}.\ ] ] each @xmath653 belongs to the space @xmath654 .
the @xmath655-image of this @xmath375-family forms a loop @xmath656}$ ] in @xmath657 .
the loop bounds a @xmath169-disk @xmath658\big)\big\}_{\{x_0 ^ 2 + x_1 ^ 2 \leq 1\}}\ ] ] in @xmath659 , which hits the subspace @xmath649 at the singleton @xmath660 .
in fact , by theorem [ th1.8 ] , @xmath664 $ ] is a generator @xmath539 of @xmath666 . in particular , the @xmath655-image of the loop @xmath667\big\}_{\theta \in [ 0 , 2\pi]}\ ] ] in @xmath654 is a generator @xmath539 of @xmath443 .
its zero set in the annulus @xmath668 $ ] ( equipped with the coordinates @xmath669 ) is a union of two loops , similar to the ones shown in fig .
9 , ( a ) .
theorem 11 from @xcite , makes an important for us claim : the @xmath670-induced map in homology @xmath671 is an isomorphism for all @xmath672 .
in particular , @xmath673 is an isomorphism .
moreover , @xmath674 is an isomorphism as well @xcite . as we proceed ,
let us keep these facts in mind .
of course it is much easier to visualize events in the @xmath48-dimensional space @xmath654 than their analogues in the infinite - dimensional @xmath657 .
this will be our next task .
it will lead us to explore the beautiful stratified geometry of the swallow tail discriminant surface .
consider the subspace @xmath675 , formed by the monic depressed ] polynomials @xmath676 . since @xmath677 is a deformation retract of @xmath654 ( @xcite ) ,
these two are homotopy equivalent .
so it is a bit easier to visualize the generator @xmath678 ( rather than @xmath679 ) , since @xmath677 is a domain in @xmath680 , the space of the coefficients @xmath681 .
the polynomials with real roots of multiplicity @xmath682 form a singular surface @xmath683 , the famous swallow tail ( see fig .
the point - polynomial @xmath684 ( the strongest singularity @xmath685 of @xmath278 ) , together with the polynomials that have one root of multiplicity @xmath51 , must be excluded from @xmath680 to form @xmath677 .
these excluded polynomials form two branches of a curve @xmath686 , whose apex @xmath685 is the origin and whose branches extend to infinity .
one branch @xmath687 of @xmath688 correspond to polynomials with the smaller simple root followed by the root of multiplicity @xmath51 ; the other branch @xmath689 of @xmath688 correspond to polynomials with the smaller root of multiplicity @xmath51 , followed by the simple root .
the curve @xmath690 , the self - intersection locus of @xmath278 , represents polynomials with two distinct roots of multiplicity @xmath169 .
it belongs to the set @xmath677 . thus @xmath691 .
therefore the generator @xmath678 is represented by an oriented loop in @xmath680 that winds once around the curve @xmath688 .
such loop @xmath692 must hit once the locus @xmath693 , formed by polynomials whose real zeros conform to the pattern @xmath342 : indeed , the curve @xmath694 is the boundary of the surface @xmath695 . traversing @xmath695 from the chamber of polynomials with @xmath48 real roots to the chamber of polynomials with @xmath169 real roots picks a normal to @xmath695 orientation .
similar rule of orientation can be applied to the stratum @xmath696 bounded by the curves @xmath697 , the stratum @xmath698 bounded by the curves @xmath699 , and the stratum @xmath700 that separates the chamber with two real roots from the chamber with no real roots at all . for any smooth loop
@xmath701 which is in general position to @xmath278 and its strata , consider the zero set @xmath702 thanks to the very definition of the target space @xmath677 , the set @xmath703 is a collection of closed curves with transversal self - intersections , no triple intersections , no self - tangencies , and no @xmath375-vertical inflections .
the cardinality of the fiber of @xmath704 does not exceed @xmath48 .
now the linking number @xmath705 equals the algebraic intersection of the loop @xmath706 with the surface @xmath695 that bounds @xmath688 .
each time the loop @xmath706 intersects the surface @xmath695 transversally in the direction of the positive normal , a point from @xmath707 is generated , and each time @xmath706 intersects @xmath695 is in the direction of the negative normal , a point from @xmath708 is generated .
in particular , the generator @xmath709 of @xmath710 has the property @xmath711 .
therefore we get @xmath712 the baby model of the formula from theorem [ th1.8 ] .
this number is an invariant of the homotopy class of the loop @xmath486 .
similarly , each time the loop @xmath706 intersects the surface @xmath700 transversally in the direction of the positive normal , a point from @xmath713 is generated , and each time @xmath706 intersects @xmath700 is in the direction of the negative normal , a point from @xmath714 is generated .
note that perhaps not any set @xmath703 is the image @xmath133 of an immersion @xmath715 for some orientable surface @xmath35 .
but the generator @xmath539 in fig . 9 , ( a ) , is .
also , if we insist that the cardinality of the @xmath375-fibers @xmath716 , we can not accommodate surfaces @xmath35 with handles .
according to remark 9.2 , to accommodate them , we need to deal with polynomials / functions of degree @xmath628 at least . *
( 111111 ) , ( 1111 ) , ( 11 ) , @xmath720 ; * ( 21111 ) , ( 12111 ) , ( 11211 ) , ( 11121 ) , ( 11112 ) , ( 211 ) , ( 121 ) , ( 112 ) , ( 2 ) ; * ( 2211 ) , ( 1221 ) , ( 1122 ) , ( 2121 ) , ( 2112 ) , ( 1212 ) , ( 22 ) ; * ( 3111 ) , ( 1311 ) , ( 1131 ) , ( 1113 ) , ( 31 ) , ( 13 ) ; * etc .
we denote by @xmath721 the set of real monic polynomials whose real divisors conform to the combinatorial pattern @xmath722 , where @xmath723 are natural numbers .
we denote by @xmath724 the @xmath725-norm of the vector @xmath719 . evidently , @xmath726 . the sets @xmath727 form a partition of the space @xmath728 . in fact , each @xmath721 is homeomorphic to an open ball of dimension @xmath729 , where @xmath730 ( @xcite ) .
the closure @xmath731 of @xmath721 in @xmath732 is an affine semi - algebraic variety . by resolving @xmath731 appropriately
, one can show that the partition @xmath727 defines a structure of a @xmath733-complex on @xmath728 , or rather , on its one - point compactification ( @xcite ) .
so we may think of @xmath721 s as being cells " ( although @xmath731 may not be homeomorphic to an infinite cone over a closed ball ) .
let @xmath734 denote the set of monic polynomials with multiple roots , the @xmath735-dimensional discriminant variety .
the first bullet lists the four @xmath628-dimensional chambers - cells in which @xmath278 divides @xmath728 .
the second bullet lists all @xmath735-dimensional strata in which @xmath278 is divided by the strata of dimension @xmath48 .
the first three bullets list the monic polynomials that form the space @xmath736 .
the third and the fourth bullets list the @xmath48-dimensional cells - strata . the forbidden locus @xmath737 is the union of strata , labeled by the combinatorial types in the fourth bullet and on .
then @xmath737 is the closure of the set @xmath738 we can orient each cell @xmath721 so that @xmath739 @xmath740 @xmath741 the operator @xmath742 in these formulas should be understood in the spirit of algebraic topology as the boundary operator on cellular chains ( and not as a topological boundary of the appropriate sets ) @xcite .
then @xmath749 belong to the locus @xmath750 . ]
therefore we get @xmath751 a version of the formula from theorem [ th1.8 ] , being applied to the loop @xmath752 .
the loop is produced by a generic with respect to @xmath110 immersion @xmath449 , such that the cardinality of the fibers of @xmath753 does not exceed @xmath628 , and by an auxiliary function @xmath754 .
these considerations are not restricted to polynomials / functions of degree @xmath628 : they apply to any even degree @xmath6 .
the application requires a deeper dive into the combinatorics of real polynomial divisors and their modifications , but the spirit is captured by the arguments that deal with degree @xmath628 ( see [ k3 ] ) .
any @xmath110-generic immersion @xmath449 also produces a well - defined element @xmath755 $ ] in the set of _ homotopy classes _ @xmath756 $ ] of maps from the trajectory graph @xmath34 to the functional space @xmath438 .
its construction is similar to the one of @xmath757 .
consider the @xmath110-generated obvious map @xmath758 ( each @xmath13-trajectory is contained in the unique @xmath110-trajectory ) .
put @xmath759 . * remark 9.3 .
* note that , for some immersions @xmath760 , the invariant @xmath480 may be different from @xmath329 , but @xmath755 $ ] may be trivial .
for example , this is the case when @xmath35 is a disk with a snake - like boundary @xmath133 with respect to @xmath110 .
however , there exist immersions @xmath22 with a nontrivial @xmath755 $ ] . for example , such is the immersion in fig . 10 , ( 1 ) . at the same time , for @xmath22 in fig . 11 , ( 3 ) , @xmath755 $ ] is trivial .
@xmath70 thus @xmath763 induces a map @xmath764 in particular , we get an element @xmath765 , where @xmath766 is a generator of @xmath762 .
this cohomology class @xmath767 is a characteristic class of the given @xmath110-generic immersion @xmath22 .
theorem [ th1.7 ] implies that if two @xmath110-generic immersions @xmath768 are such that the pull - backs @xmath769 , then @xmath770
. so the cohomology class @xmath767 is , in fact , a characteristic class of @xmath13 .
it is desirable to be able to reach this conclusion without relying on the cobordisms of curves patterns in @xmath291 with no @xmath375-vertical inflections .
based on the partial evidence , provided by the two polynomial models @xmath654 and @xmath736 , we may conjecture that the value of @xmath767 on any loop ( @xmath50-cycle ) @xmath771 equals to the linking number @xmath772 .
the validation of this conjecture requires to extend our analysis of the stratified geometry of @xmath773 to @xmath774 and to show that @xmath775 is a weak homotopy equivalence for all even @xmath6 .
both steps are realizable with the techniques developed in @xcite and in @xcite . in dimensions higher than two ,
similar considerations apply to produce characteristic classes of traversally generic flows .
they are based on computations of homology of spaces of real monic polynomials with _ restricted combinatorics _ of their real divisors .
it turns out that the topology of high - dimensional convex envelops is as intricate as the homotopy groups of spheres @xcite .
our investigation of vector flows in flatland reached its conclusion . to find out how things flow in other landsthe romances of many dimensions"(@xcite ) , the reader could consult with the references below .
alpert , h. , katz , g. , _ using simplicial volume to count multi - tangent trajectories of traversing vector fields _ , geometriae dedicata doi 10.1007/s10711 - 015 - 0104 - 6 ( arxiv:1503.02583v1 [ math.dg ] ( 9 mar 2015 ) ) .
katz , g. , _ convexity of morse stratifications and spines of 3-manifolds _ , math.gt/0611005 v1(31 oct .
katz , g. , _ stratified convexity & concavity of gradient flows on manifolds with boundary _ , applied mathematics , 2014 , 5 , 2823 - 2848 , ( also arxiv:1406.6907v1 [ mathgt ] ( 26 june , 2014 ) ) . | in this paper , we present our general results about traversing flows on manifolds with boundary in the context of the flows on surfaces with boundary . we take advantage of the relative simplicity of @xmath0-worlds to explain and popularize our approach to the morse theory on smooth manifolds with boundary ,
in which the boundary effects take the central stage . |
introduced in 1966 by a. eringnen ( see @xcite ) , micropolar equations became an important generalization of the classical navier - stokes model .
these equations take into account that fluid molecules may rotate independently of the fluid rotation .
thus , the standard navier - stokes system is complemented with another vector equation which describes the angular momentum of the particles .
if we denote the velocity field by @xmath2 and the microrotation fields by @xmath3 , then we see that @xmath4 has six degrees of freedom .
let us clearly emphasize that @xmath3 does not represent the rotation field @xmath5 derived from the velocity field ( @xmath2 ) and in most cases these vector fields differ fundamentally from each other .
this phenomenon gains an immense significance for modelling some well - known fluids , e.g. animal blood or liquid crystals ( see e.g. @xcite ) . in the microscale , when at least one dimension of the domain is only a few times larger than the size of the molecules ( e.g. blood vessels , lubricants ) , fluid motions even for isotropic fluid can differ substantially from what would follow from the computations based entirely on the navier - stokes equations ( see @xcite ) .
this behavior is caused by the dominance of the surface stresses over body forces . although not all the aspects of physical experiments have been fully explained but it is justified to assume that the surface stresses and the internal degrees of freedom of particles are the deciding factors for properties of fluid motion .
it is worth mentioning , that apart from a. eringen , other mathematicians and physicists have proposed numerous generalizations of the navier - stokes equations .
the comparison of these theories can be found in @xcite . for a short historical review
we refer the reader to ( * ? ? ?
* ch . 1 , 5 )
. in this work we plan to investigate the global existence of strong solutions to micropolar equations which are given by @xmath6 the unknowns are : the velocity field @xmath2 , the pressure @xmath7 and the microroation field @xmath3 .
the viscosity coefficients @xmath8 , @xmath9 , @xmath10 and @xmath11 are fixed and positive .
note that if @xmath9 then @xmath12 and @xmath13 get uncoupled .
therefore we can not expect better results than for the classical navier - stokes equations .
so far we have not specified the domain @xmath14 .
we assume that is has a product form @xmath15 where the constants @xmath16 and @xmath17 are positive and @xmath18 is a @xmath19-closed curve in @xmath20 .
thus , @xmath14 is a finite cylinder placed alongside the @xmath1-axis ( see figure [ fig1 ] ) . and its boundary @xmath21 . ] from practical point of view ( blood vessels , lubrication theory )
our choice is justified . from theoretical perspective , our approach is intensely focused upon search for such solutions that are close to two dimensional ( see e.g. @xcite , @xcite , @xcite ) . the solutions which are proved to exist , can be regarded as a slight perturbation of two dimensional flow along the perpendicular direction .
this perturbation will be somehow measured by @xmath22 ( see ) , which we introduce later .
we shall emphasize that since we only require the initial rate of change of the flow and microrotation , as well as the derivatives of the external data with respect to @xmath1 to be small , the flow alongside the cylinder can be large , but close to constant .
as far as the boundary condition are concerned , we use @xmath23 where @xmath24 is the unit outward vector .
the first two equations may be interpreted as tangential `` slip '' velocity being proportional to tangential stress with a factor of proportionality depending only on the curvature of @xmath18 ( see e.g. @xcite , @xcite ) . such boundary condition was already postulated in 1827 by c.m.l.h .
the third equation is clear but from the physical point view not necessarily adequate because the molecules may not move but they can rotate ( see @xcite ) . this effect is regarded in the fourth equation ( see @xcite ) .
before we present the main result of this work , we should employ some notation . by @xmath25
we denote @xmath26 where @xmath27 . a generic constant @xmath28 may change from line to line and is subscripted with appropriate symbol which indicates the dependence on the domain ( @xmath29 ) , embedding theorems ( @xmath30 ) , the poincar inequality ( @xmath31 ) and viscosity coefficients ( @xmath32 ) . to simplify the formulas we will also use @xmath33 to shorten energy estimate we introduce @xmath34 finally
, the function @xmath35 will be of particular interest .
it expresses the smallness assumption which has to be made in order to prove the existence of regular solutions on @xmath36 .
note that it contains only derivative of the external and the initial data with respect to @xmath1 .
the notation for function spaces is standard and follows ( * ? ? ?
* ch . 3 , 1.1 ) , ( * ? ? ?
* ch . 2 ,
3 ) and ( * ? ? ?
* ch . 1 , 1.1 ) : 1 .
@xmath37 , where @xmath38 , @xmath39 , is the closure of @xmath40 in the norm @xmath41 2 .
@xmath42 , where @xmath43 , is simply @xmath44 , 3 .
@xmath45 , where @xmath39 , is the closure of @xmath46 in the norm @xmath47 4 .
@xmath48 is the closure of @xmath49 in the norm @xmath50 5 .
@xmath51 , where @xmath52 and @xmath53 is a banach space , is the set of all strongly measurable functions defined on the interval @xmath54 $ ] with values in @xmath53 with finite norm defined by @xmath55 where @xmath56 and by @xmath57 for @xmath58 .
@xmath59 , where @xmath43 , is the closure of @xmath46 in the norm @xmath60
the main results of this work reads : [ t2 ] let @xmath61 and @xmath62 be sufficiently large and fixed .
suppose that @xmath63 and @xmath64 .
in addition , let the external data satisfy @xmath65 , @xmath66 , @xmath67 and @xmath68 then , for @xmath69 sufficiently small there exists a unique and regular solution to problem equipped with the boundary conditions on the interval @xmath70 .
moreover , @xmath71 the proof of this theorem is based on the result obtained in ( * ? ? ?
* theorem 1 ) : [ t1 ] let @xmath72 , @xmath73 .
suppose that @xmath74 , @xmath75 .
finally , assume that @xmath65 , @xmath66 .
then , for @xmath22 sufficiently small there exists a unique solution @xmath76 to problem supplemented with the boundary conditions such that @xmath77 and @xmath78 we see that in view of theorem [ t1 ] the extension of solutions up to the infinity with respect to time can be done in many ways .
a substantive argument would be : if the solution to on @xmath79 $ ] is regular and its estimate does not contain any constant which depend on time , then it is global .
we must emphasize that the existing terminology on the topic is not precise . according to some authors
a solution is global if the constants are time - dependent , but they do not blow up for any finite @xmath80 .
however , in such case it is more accurate to speak about large time existence instead of global existence . for we could not simply put @xmath81 , because it would lead to improper integrals and several technical difficulties .
besides , it would imply that the external data must vanish as @xmath80 goes to infinity .
hence , we adopt an alternative approach .
we consider local solution on the time interval of the form @xmath82 $ ] , where @xmath43 and @xmath83 is fixed number . starting with @xmath84
we let @xmath85 , thereby obtaining a sequence of solutions with different initial conditions @xmath86 and @xmath87 . in order to guarantee that this sequence is in fact an extension of solution from @xmath79 $ ]
, we must control the growth of the initial conditions .
additionally , if @xmath88 and @xmath89 , certain restrictions on the external data must be also imposed .
let us now briefly discuss previous results concerning the existence of global solutions of strong solutions to micropolar equations . in @xcite and @xcite
lange proved the existence and uniqueness of global and strong solutions to in hilbert spaces when the data are small enough .
his approach is based on integral equations .
this problem was also studied by sava in @xcite under homogeneous boundary conditions in the case when the body forces and moments are not present .
when the external data is present , @xmath8 is large and the data are small in comparison to @xmath8 problem with zero dirichlet boundary condition was examined by ukaszewicz in @xcite and in slightly more general framework by ortega - torres and rojas - medar in @xcite , who in contrast to ukaszewicz did not assume any decay for the external data as @xmath90 goes to the infinity .
semi - group approach was explored by yamaguchi in @xcite .
he established the global existence of strong solutions in case of small data . for further bibliographical notes
we refer the reader to @xcite , @xcite and ( * ? ? ?
* ch.3 , 5 ) . summing up : our result is established under smallness assumption not on the data itself but on its rate of change along @xmath1-variable . the external data do not vanish as @xmath90 tends to infinity but exponentially decay on time intervals of the form @xmath82 $ ]
. the boundary conditions for the velocity field and partially for the microrotation field belong to the class of slip boundary conditions .
the rest of this work consists of section [ sec4 ] , which contains some technical remarks , and section [ sec5 ] where we give estimates needed to prove theorem [ t2 ] .
in this section we present some technical facts which are essential for further considerations .
[ l1 ] let @xmath91 and @xmath92 belong to @xmath93 .
then @xmath94 it is an easy exercise .
[ l2 ] suppose that @xmath95 with either @xmath96 or @xmath97
. then @xmath98 for the proof we refer the reader @xcite , where general overdetermined elliptic systems are examined .
in particular , the case of tangent components of @xmath91 is considered .
[ lem12 ] let @xmath2 , @xmath99 and @xmath100 be given .
then the pair @xmath101 is a solution to the problem @xmath102 [ lem4 ] let @xmath103 , @xmath3 , @xmath104 and @xmath105 be given . then the function @xmath99 is solution to the problem @xmath106 proofs of lemmas [ lem12 ] and [ lem4 ] are given in @xcite .
let us notice that for the functions @xmath103 and @xmath99 the poincar inequality holds .
indeed , since @xmath107 vanishes on @xmath108 we only need to check if the integral of @xmath109 over @xmath14 equals zero .
we have @xmath110 for @xmath111 , which vanishes on @xmath108 it is also obvious . for @xmath112 we simply calculate the mean value : @xmath113 which follows from @xmath114 . [ lem110 ] suppose that @xmath103 is a solution to .
then @xmath115 consider @xmath116 introduce a partition of unity @xmath117 .
if we denote @xmath118 , then the above system becomes @xmath119 & & \text{on $ \operatorname{supp}\zeta_k\cap s_1 $ } , \\
& \bar u \cdot
n = 0 & & \text{on $ \operatorname{supp}\zeta_k\cap s_2$}. \end{aligned}\ ] ] note , that the boundary condition on @xmath120 is equal to zero which follows from @xmath13 .
there are four cases to consider : 1 . @xmath121 . : : the boundary conditions is @xmath122 . from lemma [ l2 ] we get @xmath123 2 .
@xmath124 , @xmath125 .
: : on the set @xmath126 we see that @xmath127 , whereas on @xmath128 we have @xmath122 which in particular means that @xmath127 .
next we transform the set @xmath129 into the half - space and utilize theorem 5.5 from @xcite .
it yields but in the half - space , i.e. @xmath130 3 .
@xmath124 , @xmath131 .
: : observe that @xmath132 . on the other hand we already know that @xmath133 and @xmath134 on @xmath108 ( see @xmath114 ) .
therefore we can reflect the function @xmath103 outside the cylinder following the formula @xmath135 where @xmath136 and @xmath137 .
we easily check that @xmath138 on @xmath139 and @xmath140 on @xmath141 .
then we follow case @xmath142 .
@xmath143 , @xmath131 .
: : on @xmath144 we have @xmath145 and on @xmath146 we get @xmath122 . next we map @xmath129 into the half - space and utilize theorem 5.5 from @xcite .
it yields in the half - space . summing over @xmath147 yields @xmath148 from the above inequality and using ( see ( * ? ? ?
* rem . 8.3 ) ) @xmath149 we deduce that for @xmath150 @xmath151 eventually , we demonstrate that @xmath152 .
again , from it follows that @xmath153 where we also used lemma [ l1 ] .
the boundary integral vanishes on @xmath154 due to boundary conditions @xmath13 .
on @xmath108 it can be written in a form @xmath155 and since @xmath156\vert_{s_2 } = 0 $ ] , where the last equality follows from @xmath114 , we get @xmath157 now we use the hlder and the young with @xmath158 inequalities , which results in @xmath159 and ends the proof .
[ l4 ] let @xmath72 ( see @xmath160 ) .
then for any @xmath161 we have @xmath162 this was proved in ( * ? ? ? * lemma 8.1 ) .
we begin with certain refinement of the fundamental energy estimate for the function @xmath2 and @xmath3 in the norm @xmath163 .
multiplying @xmath12 by @xmath2 and @xmath3 respectively and integrating over @xmath14 yields @xmath170 in view of @xmath160 and from @xmath171 we immediately get that @xmath172 by the application of lemma [ l1 ] we see that @xmath173 and @xmath174 in view of the boundary conditions all above boundary integrals vanish . on the right - hand side
we have @xmath175 where we applied lemma [ l1 ] .
next we see that the boundary integral vanishes on @xmath154 since @xmath176 and on @xmath108 @xmath177 = 0 $ ] holds , which follows from @xmath114 .
thus @xmath178 now we make necessary estimates . from lemma [ l2 ] for @xmath179
it follows that @xmath180 by the hlder and the young inequalities we obtain @xmath181 now we set @xmath182 , @xmath183 , @xmath184 .
since @xmath185 and @xmath186 we finally get @xmath187 with @xmath188 , the above inequality becomes @xmath189 integrating with respect to @xmath190 yields @xmath191 or equivalently @xmath192 which proofs assertion @xmath168 .
we have assumed an exponential decay with respect to time on the external data and their derivative along the axis of the cylinder .
the reason underlying this assumption follows from the necessity to control the amount of the energy supplied to the system .
if we consider the global in time existence , the supplied energy must be balanced by the rate of its lost due to friction , otherwise the system blows - up .
however , we emphasize that this exponential decay with respect to time for the external data has only a local character . if @xmath199 denotes the end of the time interval under consideration , then @xmath200 and analogously for the other external data .
the right - hand side limit can be even much larger than the left - hand side limit .
note , that alternatively we may attempt to analyze carefully the direct correspondence between the energy input and loss which is likely to result in different assumption on the external data .
however , it is beyond the scope of our study . for further reading
we refer the reader to @xcite . note also , that if we had the poincar inequality for @xmath201 and @xmath202 with respect to @xmath1 , we would be able to relax the assumption on the exponential decay of the external data and limit our considerations only to their rate of change . by application of lemma [ l1 ]
we see that @xmath207 where we used the boundary conditions @xmath208 . from the vector identity @xmath209 and lemma [ l1 ]
it follows that @xmath210 the boundary conditions @xmath114 imply that @xmath211 .
thus , @xmath212 .
since @xmath213 and @xmath214 = 0 $ ] we see that the boundary integrals vanish .
therefore @xmath215 in the same manner we get @xmath216 moving to the first term on the right - hand side we encounter a problem with the integral @xmath217 we easily see that the application of lemma [ l1 ] or integration by parts lead to a boundary integral , either @xmath218 we are not able to compute these integrals , because the boundary conditions yield insufficient information .
the first integral contains second - order derivatives of @xmath3 , which makes it impossible to estimate by the data in any suitable norm . for the second we apply trace theorem and interpolation inequality , which we shall present later .
for now we only write @xmath219 finally , we rewrite in the following form @xmath220 we shall estimate @xmath221 by application of the hlder and the young inequalities .
two first one is obvious : @xmath222 for @xmath223 more work is required . from the trace theorem
it follows that for @xmath224 @xmath225 now we use an inequality of gagliardo - nirenberg type ( see ( * ? ? ?
1 , rem . 1.2.1 ) ) @xmath226 where @xmath99 satisfies @xmath227 which allows us to set @xmath228 .
thus , by the young inequality with @xmath158 @xmath229 finally @xmath230 for the nonlinear terms we have @xmath231 and @xmath232 the two last terms are estimated in a standard way @xmath233 before we chose @xmath234 , @xmath235 , we justify the following inequality @xmath236 if we put @xmath237 ( we see that @xmath238 ) in lemma [ l1 ] then we get @xmath239 on the other hand , let now @xmath240 ( we see that @xmath96 ) in lemma [ l1 ] and use the above inequality .
then @xmath241 we set @xmath242 thus , we get from that @xmath243 where we use at the end . the next step is to estimate the @xmath244-norms on the left - hand side from below by @xmath245-norms of @xmath246 and @xmath247 . for any @xmath248 we have @xmath249 let
@xmath250 then @xmath251 where @xmath252 will be chosen later . on the other hand , in view of lemma [ l2 ] we already know that @xmath253 let @xmath254 denote @xmath255 then , combining with gives @xmath256 which is equivalent to @xmath257 we may also write @xmath258 integrating with respect to @xmath190 yields @xmath259 next we divide both sides by @xmath260 .
it gives @xmath261 consider the integral with respect to @xmath90 on the right - hand side . since @xmath262 is non - negative ,
we can write @xmath263 by assumption and assertion @xmath195 from lemma [ lem28 ] we see that @xmath264 now we chose @xmath265 .
two cases may occur : 1 . @xmath266 .
: : in this case @xmath267 so we chose such @xmath265 that it satisfies @xmath268 then @xmath269 2 .
: : now we have @xmath271 since @xmath272 we see that @xmath273 .
it is enough to set @xmath274 .
then @xmath275 and @xmath276 in both cases we have obtained an estimate of the form @xmath277 where either @xmath274 or @xmath265 satisfies . putting the above estimate into , we get from that @xmath278 in view of lemma [ l2 ] and an obvious inequality @xmath279 we can rewrite the above estimate in the form @xmath280 next observe that since @xmath281 , we have @xmath282 integrating this inequality with respect to @xmath190 and utilizing the estimate from theorem [ t1 ] we obtain @xmath283 by assumption on @xmath201 , @xmath202 , their derivative with respect to @xmath1 we see that @xmath284 thus @xmath285 and therefore for @xmath199 large enough we have @xmath286 which implies that @xmath197 where the function @xmath287 has the property that @xmath288 .
this concludes the proof .
note , that we have assumed that the @xmath291 and @xmath104 vanish on @xmath108 .
both functions appear in boundary integrals which need to be estimated .
although it is possible , but it leads to the presence of @xmath292 on the right - hand side . since @xmath30 is out of control we are unable to ensure the uniform estimates analogously as we did in lemma [ lem28 ] .
we multiply @xmath160 by @xmath293 and integrate over @xmath14 , which yields @xmath294 for the first term on the left - hand side we have @xmath295 where the boundary integrals vanish due to the boundary conditions @xmath296 .
consider next the first term on the right - hand side in .
since @xmath171 we may integrate by parts , which yields @xmath299 where we used that @xmath300 .
the last inequality above is justified in light of lemma [ lem110 ] .
setting @xmath304 and @xmath305 yields @xmath306 where in the last inequality we used @xmath307 and subsequently utilized lemma [ l2 ] . since @xmath308 we get from @xmath309 in ( * ?
* proof of lemma 8.4 ) we established @xmath310 multiplying it by @xmath311 and adding to yields @xmath312 interpolation between @xmath313 and @xmath314 shows that @xmath315 , because @xmath316 in view of we see that @xmath317 . since @xmath318
we get @xmath319 from and natural embeddings it follows @xmath320 in the same manner @xmath321 thus @xmath322 denote @xmath323 then becomes @xmath324 let @xmath325 then , the last inequality implies that @xmath326 integrating with respect to @xmath190 yields @xmath327 since @xmath328 we get that @xmath329 by assumption on @xmath100 and @xmath105 we obtain @xmath330 next we estimate the integral with respect to @xmath90 @xmath331 consider the quantity @xmath332 in view of lemma [ l4 ] we see @xmath333 to estimate the last term in we use the estimate from theorem [ t1 ] @xmath334 by assumption on @xmath201 , @xmath202 , their derivative with respect to @xmath1 we see that @xmath284 thus , @xmath335 and therefore for @xmath199 large enough we have @xmath336 which combined with implies that @xmath337 where the function @xmath338 has the property @xmath339 from the above inequality we immediately get that @xmath340 for @xmath341 the inequality yields @xmath342 which is our claim . from lemma [ lem29 ]
it follows that @xmath343by assumption @xmath344 thus , we set @xmath345 let @xmath61 , @xmath346 .
in view of theorem [ t1 ] with @xmath265 we get the existence of regular solution on the interval @xmath79 $ ] .
lemma [ lem22 ] yields the inequality @xmath347 for @xmath83 sufficiently large .
it allows us to use theorem [ t1 ] with the initial conditions @xmath348 , @xmath349 and with @xmath265 on the time interval @xmath350 $ ] . from lemma [ lem22 ]
it follows that @xmath351 which in view of previous estimate provides us with regular solution on the interval @xmath352 $ ] . reiterating this procedure yields the existence of regular solution on @xmath353 $ ] .
passing with @xmath147 up to infinity provides the global existence .
t. clopeau , a. mikeli , and r. robert , _ on the vanishing viscosity limit for the 2d incompressible navier - stokes equations with the friction type boundary conditions _ , nonlinearity * 11 * ( 1998 ) , no . 6 , 16251636 . o.a .
ladyenskaja , v.a .
solonnikov , and n.n .
uralceva , _ linear and quasilinear equations of parabolic type _
, translated from the russian by s. smith
. translations of mathematical monographs , vol . 23 , american mathematical society , providence , r.i . , 1967 . v.a .
solonnikov , _ on boundary value problems for linear parabolic systems of differential equations of general form .
* 83 * ( 1965 ) , 1184 , translation from : trudy mat .
steklov 83 ( 1965 ) , 3163 , ( russian ) .
r. temam , _ navier - stokes equations .
_ , revised ed . , studies in mathematics and its applications , vol . 2 , north - holland publishing co. , amsterdam , 1979 , theory and numerical analysis , with an appendix by f. thomasset . | the micropolar equations are a useful generalization of the classical navier - stokes model for fluids with micro - structure .
we prove the existence of global and strong solutions to these equations in cylindrical domains in @xmath0 .
we do not impose any restrictions on the magnitude of the initial and external data but we require that they can not change in the @xmath1-direction too fast . |
let @xmath0 be an integral convex polytope , that is , a convex polytope whose vertices have integer coordinates , of dimension @xmath1 .
we say that @xmath2 is a _
fano polytope _ if the origin of @xmath3 is a unique integer point in the interior of @xmath2 .
* a fano polytope is called _ terminal _ if every integer point on the boundary is a vertex . * a fano polytope
is called _ gorenstein _ if its dual polytope is integral . here , the dual polytope of a fano polytope @xmath2 is the convex polytope consisting of @xmath4 such that @xmath5 for all @xmath6 , where @xmath7 is the usual inner product of @xmath3 . *
a fano polytope is called _ @xmath8-factorial _ if it is simplicial . * a fano polytope
is called _ smooth _ if the vertices of each facet form a @xmath9-basis of @xmath10 .
in particular , smooth fano polytopes are always terminal , gorenstein and @xmath8-factorial .
fano polytopes have been studied by many people .
bro @xcite constructed the so - called sfp - algorithm which yields the complete classification list of the smooth fano polytopes of dimension @xmath1 for any given positive integer @xmath1 .
casagrande @xcite proved that the number of vertices of a gorenstein @xmath8-factorial fano polytope is at most @xmath11 when @xmath1 is even , and at most @xmath12 when @xmath1 is odd . in @xcite , nill and bro classified the gorenstein @xmath8-factorial fano polytopes of dimension @xmath1 with @xmath12 vertices .
gorenstein fano polytopes were classified when @xmath13 by kreuzer and skarke @xcite .
the study of the classification of fano polytopes of dimension three was done by kasprzyk @xcite . in this paper ,
given a finite directed graph @xmath14 , we associate a terminal gorenstein fano polytope @xmath15 , which has been already defined in @xcite when @xmath14 is a tournament graph and in @xcite when @xmath14 is a symmetric directed graph .
we study the characterization problem of directed graphs which yield smooth fano polytopes ( theorem [ main ] ) .
moreover , we show that any centrally symmetric or pseudo symmetric smooth fano polytope can be obtained from a directed graph ( theorem [ symmetricsmooth ] ) .
in addition , as an application of theorem [ main ] , we provide new examples of smooth fano polytopes whose corresponding varieties admit khler
einstein metrics ( example [ newex ] ) .
as we see in many examples in section 4 , smooth fano polytopes arising from directed graphs are very helpful to understand and useful to consider the combinatorics on smooth fano polytopes .
in this section , we construct an integral convex polytope associated with a finite directed graph and discuss the condition with which the directed graph yields a fano polytope .
most parts of this section are referred from @xcite .
let @xmath16 be a finite directed graph on the vertex set @xmath17 .
an ordered pair of vertices @xmath18 is said to be an _ arrow _ of @xmath14 and a pair without ordering @xmath19 is said to be an _ edge _ of @xmath14 .
remark that we regard @xmath20 and @xmath21 as two distinct arrows .
let @xmath22 ( resp .
@xmath23 ) denote the arrow set ( resp .
the edge set ) of @xmath14 . throughout this paper , we allow the case where both @xmath20 and @xmath21 are contained in @xmath22 and assume that @xmath14 is connected . _ let @xmath24 be the standard basis of @xmath3 . for an arrow @xmath18 in @xmath14 ,
we define @xmath25 by setting @xmath26
. moreover , we write @xmath27 for the convex hull of @xmath28 . _ _ in @xcite , @xmath15 is introduced for a tournament graph @xmath14 , which is called the _ edge polytope _ of @xmath14 , and some properties on @xmath15 are studied in ( * ? ? ?
* section 1 ) .
similarly , in ( * ? ? ?
* section 4 ) , @xmath15 is defined for a symmetric graph @xmath14 , which is denoted by @xmath29 , and called the _
symmetric edge polytope _ of @xmath14 .
_ let @xmath30 denote the hyperplane defined by the equation @xmath31 .
since each integer point of @xmath28 lies on @xmath32 , one has @xmath33 .
thus , @xmath34 .
first of all , we discuss the dimension of @xmath15 .
a sequence @xmath35 of vertices of @xmath14 is called a _ cycle _ if @xmath36 for @xmath37 . in other words ,
the edges @xmath38 form a cycle in @xmath14 .
for short , we often write @xmath39 , where @xmath40 for @xmath41 with @xmath42 .
the length of a cycle is the number of vertices ( or edges ) forming a cycle . for a cycle
@xmath39 in @xmath14 , let @xmath43 and @xmath44 .
then we may assume that @xmath45 , where @xmath46 denotes the cardinality of a finite set @xmath47 .
a cycle @xmath48 is called _ nonhomogeneous _ if @xmath49 and _ homogeneous _ if @xmath50 .
note that two arrows @xmath20 and @xmath21 form a nonhomogeneous cycle of length two .
every odd cycle is nonhomogeneous .
( here odd ( resp .
even ) cycle is a cycle of odd ( resp .
even ) length . )
the following result can be proved by similarly to ( * ? ? ?
* proposition 1.3 ) and ( * ? ? ?
* lemma 1.1 ) .
[ dim ] one has @xmath51 if and only if @xmath14 has a nonhomogeneous cycle .
we assume that @xmath14 has at least one nonhomogeneous cycle .
next , we investigate directed graphs which define fano polytopes @xmath15 .
once we know that @xmath15 is a fano polytope , one can verify that it is terminal and gorenstein ( ( * ? ? ?
* lemma 1.4 and 1.5 ) ) . the following result can be proved similarly to ( * ? ? ?
* proposition 4.2 ) and ( * ? ? ?
* lemma 1.2 ) .
[ tgf ] an integral convex polytope @xmath33 is a terminal gorenstein fano polytope of dimension @xmath52 if and only if every arrow of @xmath14 appears in a directed cycle in @xmath14 , where a cycle @xmath48 is called a directed cycle if @xmath53 is empty .
hereafter , we assume that every arrow of @xmath14 appears in a directed cycle in @xmath14 .
notice that by this condition , @xmath14 has a nonhomogeneous cycle since every directed cycle is nonhomogeneous .
_ let @xmath14 be a directed graph on the vertex set @xmath54 with the arrow set @xmath55 .
then @xmath14 , @xmath56 s and @xmath15 are as figure [ example1 ] : _ remark that the arrows @xmath57 and the arrows @xmath58 form directed cycles . in the picture of @xmath15 ,
we ignore the third coordinate of each integer point .
then the convex polytope @xmath15 of this example becomes a smooth ( in particular , terminal and gorenstein ) fano polytope of dimension two .
_ in @xcite , terminal gorenstein fano polytopes @xmath59 arising from finite partially ordered sets @xmath60 are introduced .
let @xmath61 be a partially ordered set and @xmath62 , where @xmath63 and @xmath64
. then we can regard @xmath65 as the directed graph on the vertex set @xmath66 with the arrow set @xmath67 identifying @xmath68 with @xmath69 as the same vertex , we obtain a directed graph @xmath70 on the vertex set @xmath71 . then @xmath59 is nothing but @xmath72 .
therefore , terminal gorenstein fano polytopes associated with directed graphs are a natural generalization of those defined in @xcite
. we can study these polytopes in section 2 in the similar way . _
in this section , we consider the problem of which directed graphs yield smooth fano polytopes .
first , we prove the following [ lemma ] _ ( a ) _ let @xmath73 be a cycle in @xmath14 . if there exists a facet @xmath74 of @xmath15 with @xmath75 , then @xmath76 is homogeneous .
+ _ ( b ) _ suppose @xmath77 and @xmath78 . if @xmath79 is contained in some facet @xmath74 of @xmath15 , then @xmath80 is not in @xmath74 .
\(a ) let @xmath81 , where each @xmath82 , denote the equation of the supporting hyperplane of @xmath15 which defines a facet @xmath74 .
let @xmath83 for @xmath41 , where @xmath42 .
it then follows that @xmath84 hence , @xmath76 must be homogeneous .
\(b ) we set @xmath81 as above and suppose that @xmath79 lies on this supporting hyperplane
. then one has @xmath85 .
thus , @xmath86 .
this implies that @xmath80 can not be contained in the same supporting hyperplane .
next , we prepare two notions , @xmath87 and @xmath88 .
let @xmath89 be a homogeneous cycle of length @xmath90 , where @xmath83 for @xmath41 with @xmath42 .
then there exists a unique function @xmath91 such that * @xmath92 ( resp .
@xmath93 ) if @xmath94 ( resp .
@xmath95 ) for @xmath41 ; * @xmath96 . for two distinct vertices @xmath97 and
@xmath98 of @xmath14 , the _ distance _ from @xmath97 to @xmath98 , denoted by @xmath99 , is the length of the shortest directed path in @xmath14 from @xmath97 to @xmath98 .
if there exists no directed path from @xmath97 to @xmath98 , then the distance from @xmath97 to @xmath98 is defined to be infinity .
[ main ] let @xmath14 be a connected directed graph on the vertex set @xmath100 satisfying that every arrow of @xmath14 appears in a directed cycle of @xmath14
. then the following conditions are equivalent : + _
@xmath15 is @xmath8-factorial ; @xmath15 is smooth ; + _ ( iii ) _
@xmath14 possesses no homogeneous cycle @xmath73 such that @xmath101 for all @xmath102 , where @xmath83 for @xmath41 with @xmath42 . *
( ( i ) @xmath103 ( iii ) ) * suppose that @xmath14 possesses a homogeneous cycle @xmath76 satisfying and let @xmath73 be such cycle , where @xmath83 for @xmath41 with @xmath104 .
then one has @xmath105 where @xmath106 ( resp .
@xmath107 ) if @xmath94 ( resp .
if @xmath95 ) for @xmath41 . since @xmath76 is homogeneous , one has @xmath108 , which implies that the integer points @xmath109 are not affinely independent .
let @xmath110 for @xmath41 .
in order to show that @xmath15 is not simplicial , it suffices to find a face of @xmath15 containing @xmath111 .
let @xmath112 be integers .
we write @xmath30 for the hyperplane defined by the equation @xmath81 and @xmath113 for the closed half space defined by the inequality @xmath114 .
we will show that for suitable @xmath112 , we make @xmath32 a supporting hyperplane of a face @xmath74 of @xmath15 satisfying @xmath115 and @xmath116 .
first , let @xmath117 for @xmath41 .
it then follows easily that @xmath118 lies on the hyperplane defined by the equation @xmath119 .
next , we determine @xmath120 with @xmath121 .
we set @xmath122 in particular , we have @xmath123 when there is no @xmath124 with @xmath125 . here
, we notice that one has @xmath126 where @xmath127 .
in fact , if @xmath128 , then there are @xmath124 and @xmath129 such that @xmath130 and @xmath131 . since @xmath132
, one has @xmath133 this contradicts .
now we finish determining the integers @xmath112 . since each @xmath118 lies on @xmath32 , in order to show that @xmath74 is defined by @xmath32 , it suffices to show @xmath134 let @xmath77 .
when @xmath135 and @xmath136 , then one has @xmath137 by the definition of @xmath138 . hence ,
if @xmath140 and @xmath141 , then one has @xmath142 by . hence , @xmath139 .
let @xmath143 and @xmath144 again , let @xmath77 . in each of the nine cases below , by a routine computation , we can easily show that @xmath79 is in @xmath145 .
@xmath146 for example , a routine computation of ( 1 ) is as follows .
when @xmath147 , since @xmath148 , one has @xmath149 .
when @xmath150 , since @xmath151 , one has @xmath152 .
therefore , it follows that @xmath32 is a supporting hyperplane of a face of @xmath15 which is not a simplex .
* ( ( iii ) @xmath103 ( i ) ) * suppose that @xmath15 is not simplicial , i.e. , @xmath15 possesses a facet @xmath74 which is not a simplex .
let @xmath153 be the vertices of @xmath74 , where @xmath154 , and @xmath155 the arrows with @xmath156 for @xmath157 .
we write @xmath30 for the supporting hyperplane @xmath81 defining @xmath74 .
since @xmath153 are not affinely independent , there is @xmath158 with @xmath159 satisfying @xmath160 and @xmath161 . by removing @xmath162 with @xmath163
, we may assume that @xmath164 , where @xmath165 for @xmath166 with @xmath167 .
let @xmath168 with @xmath169 and let @xmath170 denote the subgraph of @xmath14 with the arrow set @xmath171 . if @xmath172 or @xmath173 , then @xmath174 , a contradiction .
( for a graph @xmath14 and its vertex @xmath175 , @xmath176 denotes the number of edges @xmath177 in @xmath14 with @xmath178 . )
thus , @xmath179 and @xmath180 . by lemma [ lemma ] ( b ) , since @xmath181
, it can not happen that @xmath182 with @xmath183 .
moreover , since every vertex in @xmath170 is at least degree 2 , @xmath170 is not a tree .
hence @xmath170 contains a cycle , which is homogeneous by lemma [ lemma ] ( a ) .
let @xmath73 be a homogeneous cycle in @xmath14 , where @xmath83 for @xmath41 with @xmath104 .
our goal is to show that @xmath76 satisfies the inequality ( [ eq1 ] ) .
let @xmath184 be the directed shortest path in @xmath14 such that @xmath185 and @xmath186 belong to @xmath187 .
on the one hand , since @xmath188 , one has @xmath189 for @xmath190 .
hence , @xmath191 . on the other hand
, we have @xmath192 thus , @xmath193 .
therefore , the required inequality holds . *
( ( i ) @xmath103 ( ii ) ) * suppose that @xmath15 is simplicial .
then there are just @xmath52 vertices in each facet , which are linearly independent .
let @xmath194 be the @xmath195 matrix whose row vectors @xmath196 are the vertices of a facet of @xmath15 and @xmath197 the @xmath198 submatrix of @xmath194 ignoring the @xmath1th column of @xmath194 . from the theory of totally unimodular matrices
@xcite , the determinant of @xmath197 is equal to @xmath199 , which means that @xmath15 is smooth . * ( ( ii ) @xmath103 ( i ) ) * in general , every smooth fano polytope is @xmath8-factorial . for a directed graph @xmath14
, we say that @xmath14 is _ symmetric _ if @xmath78 for all @xmath77 , that is , @xmath200 .
note that when @xmath14 is symmetric , every arrow of @xmath14 is contained in a directed cycle of length two , so @xmath15 is always a terminal gorenstein fano polytope . recall that for a connected graph @xmath14 , we say that @xmath14 is _ two - connected _ if the induced subgraph with the vertex set @xmath201 is still connected for any vertex @xmath202 , and a subgraph of @xmath14 is a _ two - connected component _ of @xmath14 if it is a maximal two - connected subgraph in @xmath14 . for symmetric directed graphs ,
we obtain the following [ noeven ] assume that @xmath14 is a connected symmetric directed graph .
then the following conditions are equivalent : + _
@xmath15 is @xmath8-factorial ; @xmath15 is smooth ; @xmath14 contains no even cycle ; + _ ( iv ) _ every two - connected component of @xmath14 is either one edge or an odd cycle . * ( ( i ) @xmath203 ( ii ) ) * this equivalence follows from theorem [ main ] . * ( ( i ) @xmath103 ( iii ) ) * suppose that @xmath14 possesses an even cycle @xmath76 in @xmath14 of length @xmath204 .
let @xmath205 be a cycle , where @xmath83 for @xmath206 with @xmath207 .
since @xmath14 is symmetric , there are arrows of @xmath14 @xmath208 we define @xmath209 by setting @xmath210 then one has @xmath211 thus , @xmath212 are not affinely independent .
hence , we may show that there is a face @xmath74 of @xmath15 with @xmath213 .
now , we have @xmath214 and @xmath215 for @xmath41 .
thus , @xmath212 lie on the hyperplane @xmath30 defined by the equation @xmath216 .
in addition , it is clear that @xmath56 is contained in @xmath113 for any arrow @xmath217 of @xmath14 .
hence , @xmath32 is a supporting hyperplane defining a face @xmath74 of @xmath15 with @xmath213 .
therefore , @xmath15 is not simplicial .
* ( ( iii ) @xmath103 ( iv ) ) * we prove this implication by elementary graph theory .
suppose that there is a two - connected component of @xmath14 which is neither one edge nor an odd cycle .
let @xmath170 be such two - connected subgraph of @xmath14 .
now , an arbitrary two - connected graph with at least three vertices can be obtained by the following method : starting from a cycle and repeatedly appending an @xmath218-path to a graph @xmath218 that has been already constructed .
( consult , e.g. , @xcite . ) since @xmath170 is not one edge , @xmath170 has at least three vertices .
thus , there is one cycle @xmath219 and @xmath220 paths @xmath221 such that @xmath222 . since @xmath170 is not an odd cycle
, one has @xmath223 , where @xmath219 is an even cycle , or @xmath224 .
suppose that @xmath225 and @xmath219 is an odd cycle .
let @xmath175 and @xmath226 be distinct two vertices of @xmath219 which are intersected with @xmath227 .
then there are two paths in @xmath219 from @xmath175 to @xmath226 .
since @xmath219 is odd , the parities of the lengths of such two paths are different . by attaching the path @xmath227 to one or another of such two paths
, we can construct an even cycle .
therefore , there exists an even cycle . *
( ( iv ) @xmath103 ( i ) ) * suppose that each two - connected component of @xmath14 is either one edge or an odd cycle . then there is no homogeneous cycle in @xmath14 . hence , by theorem [ main ] , @xmath15 is simplicial .
in this section , we show that every pseudo symmetric smooth fano polytope can be obtained from some directed graph with no even cycle .
this fact includes the case of centrally symmetric smooth fano polytopes .
let @xmath0 be a fano polytope .
* we call @xmath2 _ centrally symmetric _ if @xmath228 .
* we call @xmath2 _ pseudo symmetric _ if there is a facet @xmath74 of @xmath2 such that @xmath229 is also a facet of @xmath2 . by the definition ,
every centrally symmetric fano polytope is pseudo symmetric . *
a _ del pezzo polytope _ of dimension @xmath230 is a convex polytope @xmath231 whose corresponding variety is so - called a _ del pezzo variety _ @xmath232 . in particular , del pezzo polytopes are centrally symmetric smooth fano polytopes . *
a _ pseudo del pezzo polytope _ of dimension @xmath230 is a convex polytope @xmath233 whose corresponding variety is so - called a _ pseudo del pezzo variety _ @xmath234 .
in particular , pseudo del pezzo polytopes are pseudo symmetric smooth fano polytopes .
* let us say that @xmath2 _ splits _ into @xmath235 and @xmath236 if @xmath2 is the convex hull of two fano polytopes @xmath237 and @xmath238 with @xmath239 , i.e. , by renumbering coordinates , we have @xmath240 there is a well - known fact on the characterization of centrally symmetric or pseudo symmetric smooth fano polytopes . [ delpezzo ]
any centrally symmetric smooth fano polytope splits into copies of the closed interval @xmath241 $ ] or a del pezzo polytope . [ pdelpezzo ]
any pseudo symmetric smooth fano polytope splits into copies of the closed interval @xmath241 $ ] or a del pezzo polytope or a pseudo del pezzo polytope .
somewhat surprisingly , we also give the complete characterization of centrally symmetric or pseudo symmetric smooth fano polytopes by means of directed graphs .
in fact , [ symmetricsmooth ] _ ( a ) _ any centrally symmetric smooth fano polytope can be obtained from a symmetric directed graph with no even cycle .
+ _ ( b ) _ any pseudo symmetric smooth fano polytope can be obtained from a directed graph with no even cycle .
first , we prove ( b ) .
let @xmath2 be an arbitrary pseudo symmetric smooth fano polytope of dimension @xmath1 . by theorem [ pdelpezzo ] , @xmath2 splits into @xmath242 which are copies of @xmath241 $ ] or a del pezzo polytope or a pseudo del pezzo polytope .
let @xmath243 be del pezzo polytopes , @xmath244 pseudo del pezzo polytopes and @xmath245 the closed intervals @xmath241 $ ] . then the following arguments easily follow .
* let , say , @xmath235 be a del pezzo polytope of dimension @xmath246 and @xmath247 a symmetric directed graph with its arrow set @xmath248 then @xmath247 is an odd cycle , i.e. , there is no even cycle , so @xmath249 is smooth by corollary [ noeven ] and we can check that @xmath249 is unimodular equivalent to @xmath235 . *
let , say , @xmath250 be a pseudo del pezzo polytope of dimension @xmath246 and @xmath251 a directed graph with its arrow set @xmath252 i.e. , we miss one arrow from @xmath247 .
then we can also check that @xmath253 is unimodular equivalent to @xmath250 . * a directed graph consisting of only one symmetric edge yields the smooth fano polytope of dimension one , which is nothing but the closed interval @xmath241 $ ] . by connecting the above graphs with one vertex , we obtain the directed graph with no even cycle which yields the required smooth fano polytope @xmath2 .
moreover , del pezzo polytopes and the closed interval @xmath241 $ ] are constructed by symmetric directed graphs .
therefore , in the similar way to the above construction , by theorem [ delpezzo ] , we can also find the symmetric directed graph @xmath14 with no even cycle such that @xmath15 is unimodular equivalent to @xmath2 for any centrally symmetric smooth fano polytope @xmath2 , also proving ( a ) .
in this section , we provide examples of smooth fano polytopes arising from directed graphs .
_ let @xmath14 be a directed cycle of length @xmath69 .
then @xmath15 is a smooth fano polytope whose corresponding toric fano variety is a @xmath1-dimensional complex projective space @xmath254 .
the left - hand side ( resp .
right - hand side ) of the graph in figure [ example_proj ] yields a smooth fano polytope which corresponds to @xmath255 ( resp .
@xmath256 ) . here
each two - connected component of a directed graph corresponds to each direct factor of the corresponding toric fano variety .
_ _ ( a ) when @xmath14 is a symmetric directed graph without even cycle , @xmath15 is a smooth fano polytope whose corresponding toric fano variety is a direct product of copies of @xmath258 or del pezzo variety @xmath232 .
( see section 3 . ) for example , the left - hand side ( resp .
right - hand side ) of the graph in figure [ example_sym ] yields a smooth fano polytope which corresponds to @xmath259 ( resp .
@xmath260 ) .
_ _ in @xcite , the definition of a so - called _ symmetric _ smooth toric fano variety is given , which is important from the viewpoint whether smooth toric fano variety admits a khler
einstein metric , and some examples of symmetric smooth toric fano varieties are provided in ( * ? ? ?
* example 4.2 4.4 ) .
note that smooth toric fano varieties corresponding to centrally symmetric smooth fano polytopes and direct products of copies of complex projective spaces are symmetric .
_ let @xmath265 be a positive integer and @xmath247 a directed graph with its arrow set @xmath266 then @xmath249 is a smooth fano polytope of dimension @xmath267 which corresponds to the example of the case with @xmath268 described in ( * ? ? ?
* example 4.2 ) .
let @xmath269 be a directed graph with its arrow set @xmath270 then @xmath271 is a smooth fano polytope of dimension @xmath272 which is the example of the case with @xmath268 described in ( * ? ? ?
* example 4.3 ) .
_ by generalizing the above graphs @xmath247 and @xmath269 , we obtain a new family of symmetric smooth toric fano varieties . for a positive integer @xmath265 and nonnegative integers
@xmath273 with @xmath274 , let @xmath275 denote the directed graph on the vertex set @xmath276 with the arrow set @xmath277 where @xmath278 notice that @xmath279 and @xmath280 . it then follows from theorem [ main ] that @xmath281 is a smooth fano polytope of dimension @xmath282 if and only if the integers @xmath283 satisfy one of the following conditions : _ | in this paper , we consider terminal gorenstein fano polytopes arising from finite directed graphs and study the problem of deciding directed graphs which yield smooth fano polytopes .
we show that any centrally symmetric or pseudo symmetric smooth fano polytopes can be obtained from directed graphs .
moreover , by using directed graphs , we provide new examples of smooth fano polytopes whose corresponding varieties admit khler
einstein metrics . |
computing distances between shapes of open or closed curves is of interest in many applications , from medical imaging to radar signal processing , as soon as one wants to compare , classify or statistically analyze trajectories or contours of objects .
while the shape of an organ or the trajectory of an object on a short distance can be modeled by a curve in the plane @xmath7 or in the ambient space @xmath8 , some applications provide curves in an intrinsically non - flat space .
simple examples in positive curvature include trajectories on the sphere where points represent positions on the earth , and a space of interest in negative curvature is the hyperbolic half plane , which as we will explain later coincides with the statistical manifold of gaussian densities .
we are motivated by the study of curves in the latter for signal processing purposes , however our framework is more general .
here we consider open oriented curves in a riemannian manifold @xmath0 , more precisely the space of smooth immersions @xmath9 \rightarrow m$ ] , @xmath10,m ) = \ { c\in c^\infty([0,1],m),\ , c'(t ) \neq 0 \,\ , \forall t\in [ 0,1 ] \}.\ ] ] to compare or average elements of this space , one way to proceed is to equip @xmath5 with a riemannian structure , that is to locally define a scalar product @xmath11 on its tangent space @xmath12 .
a property that is usually required of this metric is reparametrization invariance , that is that the metric be the same at all points of @xmath5 representing curves that are identical modulo reparametrization .
two curves are identical modulo reparametrization when they pass through the same points of @xmath0 but at different speeds .
reparametrizations are represented by increasing diffeomorphisms @xmath13 \rightarrow [ 0,1]$ ] ( so that they preserve the end points of the curves ) , and their set is denoted by @xmath14)$ ] .
elements @xmath15 of the tangent space in @xmath16 are infinitesimal deformations of @xmath17 and can be seen as vector fields along the curve @xmath17 in @xmath0 ( this results from the so called `` exponential law '' for smooth functions , see e.g. @xcite ) .
the riemannian metric @xmath11 is reparametrization invariant if the action of @xmath14)$ ] is isometric for @xmath11 @xmath18 for any @xmath19 , @xmath20 and @xmath21)$ ] .
this is often called the _ equivariance property _ , and it guarantees that the induced distance between two curves @xmath22 and @xmath23 does not change if we reparametrize them by the same diffeomorphism @xmath24 @xmath25 what s more , a reparametrization invariant metric on the space of curves induces a riemannian structure on the `` shape space '' , where the space of reparametrizations is quotiented out .
a shape can be seen as the equivalence class of all the curves that are identical modulo a change of parameterization , and the shape space as the associated quotient space @xmath26,m)/\text{diff}^+([0,1]).\ ] ] while the space of immersions is a submanifold of the frchet manifold @xmath27,m)$ ] ( see @xcite , theorem 10.4 . ) , the shape space is not a manifold and therefore the fiber bundle structure we discuss next is to be understood formally .
we get a principal bundle structure @xmath28 , which induces a decomposition of the tangent bundle @xmath29 into a vertical subspace @xmath30 consisting of all vectors tangent to the fibers of @xmath3 over @xmath31 , and a horizontal subspace @xmath32 defined as the orthogonal complement of @xmath33 according to the metric @xmath11 that we put on @xmath3 . if @xmath11 verifies the equivariance property , then it induces a riemannian metric @xmath34 on the shape space , for which the geodesics are the projected horizontal geodesics of @xmath5 for @xmath11 .
the geodesic distances @xmath35 on @xmath5 and @xmath36 on @xmath37 are then linked by @xmath38 , [ c_1 ] \right ) = \inf \left\{\ , d\left(c_0 , c_1\circ \phi \right ) \ , | \ , \ , \phi \in \text{diff}^+([0,1 ] ) \ , \right\},\ ] ] where @xmath39 $ ] and @xmath40 $ ] denote the shapes of two given curves @xmath22 and @xmath23 , and @xmath36 verifies the stronger property @xmath41 for any reparametrizations @xmath42)$ ] .
this motivates the choice of a reparametrization invariant metric on @xmath5 .
riemannian metrics on the space of curves lying in the euclidean space @xmath43 , and especially closed curves @xmath44 ( @xmath45 is the circle ) , have been widely studied ( @xcite , @xcite , @xcite , @xcite ) .
the most natural candidate for a reparametrization invariant metric is the @xmath4-metric with integration over arc length @xmath46 @xmath47 but michor and mumford have shown in @xcite that the induced metric on the shape space always vanishes . this has motivated the study of sobolev - type metrics ( @xcite , @xcite , @xcite , @xcite ) , where higher order derivatives are introduced .
local existence and uniqueness of geodesics for first and second - order sobolev metrics on the space of closed plane curves were shown in @xcite and completion results were given in @xcite .
one first - order sobolev metric where different weights are given to the tangential and normal parts of the derivative has proved particularly interesting for the applications ( @xcite , @xcite ) @xmath48 in that case @xmath17 is a curve in @xmath43 , @xmath49 denotes the euclidean metric on @xmath50 , @xmath51 is the derivation of @xmath52 according to arc length , @xmath53 is the projection of @xmath54 on the unit speed vector field @xmath55 , and @xmath56 .
this metric belongs to the class of so - called elastic metrics , defined for any weights @xmath57 as @xmath58 the parameters @xmath59 and @xmath60 respectively control the degree of bending and stretching of the curve .
srivastava et al .
introduced in @xcite a convenient framework to study the case where @xmath61 and @xmath62 by showing that metric could be obtained by pullback of the @xmath4-metric via a simple transformation @xmath63 called the square root velocity function ( srvf ) , which associates to each curve its velocity renormalized by the square root of its norm .
a similar idea had been previously introduced in @xcite and then used in @xcite , where a sobolev metric is also mapped to an @xmath4-metric .
the general elastic metric @xmath64 with weights @xmath59 and @xmath60 satisfying @xmath65 can also be studied using a generalization of the srvf @xcite , however the formulas get rather involved .
the srv framework can be extended to curves in a general manifold by using parallel transport , in a way which allows us to move the computations to the tangent space to the origin of one of the two curves under comparison @xcite , @xcite , @xcite . in @xcite
this is done for the general elastic metric @xmath64 , but no riemannian framework is given . in @xcite ,
a riemannian framework is given for the case @xmath61 , @xmath66 , and the geodesic equations are derived . in this paper
we also restrict to this particular choice of coefficients @xmath59 and @xmath60 for simplicity , but we propose another generalization of the srv framework to manifold - valued curves . instead of encoding the information of each curve within a tangent space at a single point as in @xcite and @xcite , the distance is computed in the manifold itself which enables us to be more directly dependent on its geometry .
intuitively , the data of each curve is no longer concentrated at any one point , and so the energy of the deformation between two curves takes into account the curvature of the manifold along the entire `` deformation surface '' , not just along the path traversed by the starting point of the curve . in the following section ,
we introduce our metric as the pullback of a quite natural metric on the tangent bundle @xmath12 , and show that it induces a fiber bundle structure over the manifold @xmath0 seen as the set of starting points of the curves . in section @xmath67 ,
we give the induced geodesic distance and highlight the difference with respect to the distance introduced in @xcite . in section 4 , we give the geodesic equations associated to our metric and exploit them to build the exponential map .
geodesics of the space of curves can then be computed using geodesic shooting . to this end
, we describe the jacobi fields on @xmath5 .
we test these algorithms on curves lying in the hyperbolic half - plane , a choice that we motivate in section 5 .
finally , in the setting of spectral analysis of locally stationary radar signals , we compute the mean of curves in that space constructed from radar data .
let @xmath9 \rightarrow m$ ] be a curve in @xmath0 and @xmath15 two infinitesimal deformations .
we consider the following first - order sobolev metric on @xmath5 @xmath68 where we integrate according to arc length @xmath69 , @xmath70 and @xmath71 respectively denote the riemannian metric and the associated levi - civita connection of the manifold @xmath0 , @xmath72 is the covariant derivative of @xmath52 according to arc length , and @xmath73 and @xmath74 are its tangential and normal components respectively , with the notation @xmath75 .
if @xmath0 is a flat euclidean space , we obtain the metric studied in @xcite , with an added term involving the origins . without this extra term , the bilinear form @xmath11 is not definite since it vanishes if @xmath52 or @xmath76 is covariantly constant along @xmath17 .
here we show that @xmath11 can be obtained as the pullback of a very natural metric @xmath77 on the tangent bundle @xmath12 .
we consider the square root velocity function ( srvf , introduced in @xcite ) on the space of curves in @xmath0 , @xmath78 where @xmath79 is the norm associated to the riemannian metric on @xmath0 . in order to define @xmath77 , we introduce the following projections from @xmath80 to @xmath81 .
let @xmath82 and @xmath83 be a curve in @xmath81 that passes through @xmath84 at time @xmath85 at speed @xmath86 .
then we define the vertical and horizontal projections @xmath87 the horizontal and vertical projections live in the tangent bundle @xmath88 and are not to be confused with the horizontal and vertical parts which live in the double tangent bundle @xmath80 and will be denoted by @xmath89 , @xmath90 .
furthermore , let us point out that the horizontal projection is simply the differential of the natural projection @xmath91 , and that according to these definitions , a very natural metric on the tangent bundle @xmath88 , the sasaki metric ( @xcite , @xcite ) , can be written @xmath92 where @xmath49 denotes the riemannian metric on @xmath0 .
now we can define the metric that we put on @xmath93 .
let us consider @xmath94 and @xmath95 .
we define @xmath96 where @xmath97 and @xmath98 are the horizontal and vertical projections of @xmath99 for all @xmath100
. then we have the following result .
the metric @xmath11 on the space of curves @xmath5 can be obtained as pullback of the metric @xmath77 by the square root velocity function @xmath63 , that is @xmath101 for any curve @xmath19 , and vector fields @xmath20 along @xmath17 . for any @xmath102
$ ] , we have @xmath103 and @xmath104 . to prove this proposition
, we just need to compute the latter .
let @xmath105 be a curve in @xmath3 such that @xmath106 and @xmath107 . here and
in all the paper we use the notations @xmath108 and @xmath109 , as well as @xmath110 and @xmath111 .
then @xmath112 where in the last step we use again the inversion @xmath113 .
this choice of metric induces two fiber bundle structures .
while the second one is an actual fiber bundle structure between two manifolds , the first structure is over the shape space which , as discussed in the introduction , is not a manifold , and so it should be understood formally .
note that we could obtain a manifold structure by restricting ourselves to so - called `` free '' immersions , that is elements of @xmath5 on which the diffeomorphism group acts freely , see @xcite .
just as in the planar case , the fact that the square root velocity function @xmath63 satisfies @xmath114 for all @xmath115 , @xmath20 and @xmath21)$ ] , guarantees that the integral part of @xmath11 is reparametrization invariant .
remembering that the reparametrizations @xmath116)$ ] preserve the origins of the curves , we notice that @xmath11 is constant along the fibers and verifies the equivariance property .
we then have a formal principal bundle structure over the shape space @xmath117,m ) \rightarrow \mathcal{s}=\mathcal{m}/ \text{diff}^+([0,1]).\ ] ] which induces a decomposition @xmath118 .
there exists a riemannian metric @xmath34 on the shape space @xmath31 such that @xmath119 is ( formally ) a riemannian submersion from @xmath120 to @xmath121 @xmath122 where @xmath123 and @xmath124 are the horizontal parts of @xmath52 and @xmath76 , as well as the horizontal lifts of @xmath125 and @xmath126 , respectively .
this expression does in fact define @xmath34 in the sense that it does not depend on the choice of the representatives @xmath17 , @xmath52 and @xmath76 . for more details , see the theory of riemannian submersions and g - manifolds in @xcite .
the special role played by the starting point in the metric @xmath11 induces another fiber bundle structure , where the base space is the manifold @xmath0 , seen as the set of starting points of the curves , and the fibers are composed of the curves which have the same origin .
the projection is then @xmath127 it induces another decomposition of the tangent bundle in vertical and horizontal bundles @xmath128 we have the usual decomposition @xmath129 , the horizontal bundle @xmath130 consists of parallel vector fields along @xmath17 , and @xmath131 is a riemannian submersion for @xmath120 and @xmath132 .
let @xmath52 be a tangent vector .
consider @xmath133 the parallel vector field along @xmath17 with initial value @xmath134 .
it is a horizontal vector , since its vanishing covariant derivative along @xmath17 assures that for any vertical vector @xmath135 we have @xmath136 .
the difference @xmath137 between those two vectors has initial value @xmath85 and so it is a vertical vector , which gives a decomposition of @xmath52 into a horizontal vector and a vertical vector . the definition of @xmath138 as the orthogonal complement of @xmath139 guaranties that their sum is direct . now
if @xmath76 is another tangent vector , then the scalar product between their horizontal parts is @xmath140 and this completes the proof .
here we give an expression of the geodesic distance induced by the metric @xmath11 .
we show that it can be written similarly to the product distance given in @xcite and @xcite , with an added curvature term .
let us consider two curves @xmath141 , and a path of curves @xmath142 linking them in @xmath3 @xmath143 for all @xmath144 $ ] .
we denote by @xmath145 the image of this path of curves by the srvf @xmath63 .
note that @xmath146 is a vector field along the surface @xmath17 in @xmath0 .
let now @xmath147 be the `` raising '' of @xmath146 in the tangent space @xmath148 defined by @xmath149 where we denote by @xmath150 the parallel transport along a curve @xmath151 from @xmath152 to @xmath153 .
notice that @xmath147 is a surface in a vector space , as illustrated in figure [ fig : dist ] .
lastly , we introduce a vector field @xmath154 in @xmath0 , which parallel translates @xmath155 along @xmath156 to its origin , then along @xmath157 and back down again , as shown in figure [ fig : dist ] .
more precisely @xmath158 for all @xmath159 . that way the quantity @xmath160 measures the holonomy along the rectangle of infinitesimal width shown in figure [ fig : dist ] .
[ distance1 ] with the above notations , the geodesic distance induced by the riemannian metric @xmath11 between two curves @xmath22 and @xmath23 on the space @xmath161,m)$ ] of parameterized curves is given by @xmath162 where @xmath163 is the square root velocity representation of the curve @xmath17 and the norm is the one associated to the riemannian metric on @xmath0 .
it can also be written as a function of the `` raising '' @xmath147 of @xmath146 in the tangent space @xmath164 defined by , @xmath165 where @xmath166 is a curvature term measuring the holonomy along a rectangle of infinitesimal width @xmath167 if @xmath168 denotes the curvature tensor of the manifold @xmath0 and @xmath169 is defined by .
and @xmath23 in the space of curves @xmath3.,width=309 ] the second expression highlights the difference with respect to the distance given in @xcite and @xcite . in the first term under the square root
we can see the velocity vector of the curve @xmath157 linking the two origins , and in the second the velocity vector of the curve @xmath147 linking the tsrvf - images of the curves transported square root velocity function , as introduced by su et al . in @xcite .
if instead we equip the tangent bundle @xmath2 with the metric @xmath170 for @xmath171 and @xmath172 , then the curvature term @xmath166 vanishes and the geodesic distance on @xmath3 becomes @xmath173 which corresponds exactly to the geodesic distance introduced by zhang et al . in @xcite on the space @xmath174,t_pm)$ ] .
the difference between the two distances and resides in the curvature term @xmath166 , which measures the holonomy along the rectangle of infinitesimal width shown in figure [ fig : dist ] , and arises from the fact that in the first one , we compute the distance in the manifold , whereas in the second , it is computed in the tangent space to one of the origins of the curves .
therefore , the first one takes more directly into account the `` relief '' of the manifold between the two curves under comparison .
for example , if there is a `` bump '' between two curves in an otherwise relatively flat space , the second distance might not see it , whereas the first one should thanks to the curvature term .
let us briefly consider the flat case : if the manifold @xmath0 is flat , e.g. @xmath175 , then the two distances and coincide .
if two curves @xmath22 and @xmath23 in @xmath43 have the same starting point @xmath176 , the first summand under the square root vanishes and the distance becomes the @xmath4-distance between the two srv representations @xmath177 and @xmath178 .
if two @xmath43-valued curves differ only by a translation , then the distance is simply the distance between their origins .
note that this distance is only local in general . since @xmath11 is defined by pullback of @xmath77 by the srvf @xmath63 , we know that the lengths of @xmath17 in @xmath3 and of @xmath179 in @xmath2 are equal and so that @xmath180 with @xmath181 to obtain the second expression of this distance we need to express @xmath182 as a function of the derivative @xmath183 .
let us fix @xmath144 $ ] , and consider the vector field @xmath184 along the surface @xmath185 that is parallel along all curves @xmath156 and takes value @xmath186 in @xmath187 for any @xmath188 $ ] , that is @xmath189 for all @xmath190 $ ] and @xmath191 $ ] .
with this definition we have @xmath192 . since @xmath193 is a vector field along @xmath194 ,
we can write @xmath195 noticing that we additionally have @xmath196 for all @xmath197 $ ] , and using @xmath198 , the covariant derivative in @xmath199 can be written @xmath200 now let us fix @xmath190 $ ] as well . notice that the vector field @xmath169 defined above as @xmath201 verifies @xmath202,\label{nablaomega1}\\ \nabla_a\omega^{s , t}(a,0)&= & 0 \quad \forall a \in [ 0,1].\label{nablaomega2}\end{aligned}\ ] ] note that unlike @xmath184 , we do _ not _ have @xmath203 because @xmath204 is only true for @xmath205 .
using equations and we get @xmath206 which is the same integral as the one in .
finally , since @xmath207 we obtain @xmath208 which gives equation and completes the proof .
to be able to compute the distance given by between two curves , we first need to compute the optimal deformation @xmath209 from one to the other .
that is , we need to characterize the geodesics of @xmath5 for our metric . in order to do so , taking inspiration from @xcite , we use the variational principle . in what follows ,
we use the lighter notation @xmath210 to denote the parallel transport of a vector @xmath211 along a curve @xmath17 from @xmath212 to @xmath213 , when there is no ambiguity on the choice of @xmath17 .
we also denote by @xmath214 the tangential component of any vector field @xmath215 along a curve @xmath17 , that is its projection on the unit speed vector field @xmath75 .
let @xmath216\ni s \mapsto c(s,\cdot)\in\mathcal m$ ] be a path of curves .
it is a geodesic of @xmath5 if and only if it verifies the following equations @xmath217 where @xmath218 is the srv representation of @xmath17 , the vector field @xmath219 is given by @xmath220 and @xmath221 with @xmath222 , is the tangential component of @xmath219 .
the path @xmath17 is a geodesic if and only if it is a critical point of the energy functional @xmath223,\mathcal m ) \rightarrow
\mathbb r_+$ ] , @xmath224 let @xmath225 , @xmath226 , be a proper variation of the path @xmath209 , meaning that it coincides with @xmath17 in @xmath227 , and it preserves its end points @xmath228 then @xmath17 is a geodesic of @xmath5 if and only if @xmath229 for any proper variation @xmath230 .
if we denote by @xmath231 , for @xmath232 , the energy of a proper variation @xmath230 , then we have @xmath233 where @xmath234 is the srv representation of @xmath230 .
its derivative is given by @xmath235 considering that the variation preserves the end points , integration by parts gives @xmath236 and so we obtain @xmath237 this quantity has to vanish in @xmath227 for all proper variations @xmath230 @xmath238 we can not yield any conclusions at this point , because @xmath239 and @xmath240 can not be chosen independently , since @xmath241 is not any vector field along @xmath230 but its image via the square root velocity function . computing the covariant derivative of @xmath242 according to @xmath59 gives @xmath243 , and projecting both sides on @xmath244 results in @xmath245 .
we deduce @xmath246 and since @xmath247 , we can express the variation @xmath248 as follows @xmath249 inserting this expression in the derivative of the energy we obtain the following , where we omit to write that the variations @xmath230 and @xmath241 are always taken in @xmath227 for the sake of readability , @xmath250 with the previously given definition of @xmath219 . since the variations @xmath251 and @xmath240 can be chosen independently and take any value for all @xmath252 and all @xmath100 , we obtain the desired equations . now that we have the geodesic equations , we are able to describe an algorithm which allows us to compute the geodesic @xmath209 starting from a point @xmath16 at speed @xmath253
. this amounts to finding the optimal deformation of the curve @xmath17 in the direction of the vector field @xmath254 according to our metric .
we initialize this path @xmath209 by setting @xmath255 and @xmath256 , and we propagate it using iterations of fixed step @xmath257 .
the aim is , given @xmath156 and @xmath258 , to deduce @xmath259 and @xmath260 .
the first is obtained by following the exponential map on the manifold @xmath0
@xmath261 for all @xmath102 $ ] and the second requires the computation of the variation @xmath262 @xmath263^{s , s+\epsilon},\ ] ] for all @xmath102 $ ] where once again , we use the notation @xmath264 for the parallel transport of a vector field @xmath265 along a curve @xmath266 in @xmath0 . if we assume that at time @xmath252 we have @xmath156 and @xmath258 at our disposal
, then we can estimate @xmath267 and @xmath268 , and deduce @xmath269 as well as @xmath270 using the fact that @xmath113 .
the variation @xmath271 can then be computed in the following way @xmath272^{\tau , t } \mathrm d\tau\ ] ] for all @xmath102 $ ] , where @xmath273 is given by equation , the second order variation @xmath274 is given by @xmath275 and @xmath276 can be computed via equation .
[ alg : expmap ] input : @xmath22 , @xmath254 .
+ initialization : set @xmath277 and @xmath278 for all @xmath102 $ ] .
+ heredity : for @xmath279 , set @xmath280 with @xmath281 and 1 .
compute for all @xmath100 @xmath282 where @xmath283 denotes the inverse of the exponential map on @xmath0 , and compute @xmath284 and @xmath285 using equation .
2 . compute @xmath286 and @xmath287 and deduce @xmath288 using equation for all @xmath102 $ ] .
initialize @xmath289 and compute @xmath290 for all @xmath100 using equation .
4 . finally ,
for all @xmath102 $ ] , set @xmath291^{s , s+\epsilon},\end{aligned}\ ] ] where @xmath292 is the exponential map on the manifold @xmath0 .
output : @xmath293 .
the last step needed to compute the optimal deformation between two curves @xmath22 and @xmath23 is to find the appropriate initial speed @xmath254 , that is the one that will connect @xmath22 to @xmath23 .
since we do not have an explicit expression for this appropriate initial speed , we compute it iteratively using geodesic shooting .
the aim of geodesic shooting is to compute the geodesic linking two points @xmath294 and @xmath295 of a manifold @xmath296 , knowing the exponential map @xmath297 .
more precisely , the goal is to iteratively find the initial speed @xmath298 such that @xmath299 an initial speed vector @xmath300 is chosen , and is iteratively updated after evaluating the gap between the point @xmath301 obtained by taking the exponential map at point @xmath294 in @xmath254 that is , by `` shooting '' from @xmath294 in the direction @xmath254 and the target point @xmath295 .
assuming that the current point @xmath176 is `` not too far '' from the target point @xmath295 , and that there exists a geodesic linking @xmath294 to @xmath295 , we can consider that the gap between @xmath176 and @xmath295 is the extremity of a jacobi field @xmath302 \rightarrow \mathcal n$ ] in the sense that it measures the variation between the geodesics @xmath303 and @xmath304 .
since both geodesics start at @xmath294 , this jacobi field has value @xmath305 in @xmath85 .
then , the current speed vector can be corrected by @xmath306 as shown in figure [ fig : geodshoot ] .
let us briefly explain why . if @xmath307 , @xmath308 $ ] , is a family of geodesics starting from the same point @xmath294 at different speeds @xmath309 , i.e. @xmath310 , and @xmath311 , @xmath188 $ ] measures the way that these geodesics spread out , then we have @xmath312 in the context of geodesic shooting between two curves @xmath22 and @xmath23 in @xmath5 , the speed vector @xmath254 can be initialized using the @xmath4 logarithm map , the inverse of the exponential map for the @xmath4-metric ( these maps are simply obtained by post - composition of mappings with the finite - dimensional maps @xmath292 and @xmath313 ) . that is , we set @xmath314 the @xmath4 logarithm map also allows us to approximate the gap between the current point and the target point .
this amounts to minimizing the functional @xmath315 .
we summarize as follows . ,
width=264 ] [ alg : geodshoot ] input : @xmath316 .
+ initialization : set @xmath317 .
fix a threshold @xmath318 . 1 .
compute @xmath319 with algorithm [ alg : expmap ] .
2 . estimate the gap @xmath320 .
3 . if @xmath321 , set @xmath322 and @xmath323 where @xmath324 is computed using algorithm [ alg : jacobi ] , and go back to the first step .
+ else , stop .
output : @xmath17 approximation of the geodesic linking @xmath22 and @xmath23 .
the function @xmath24 associates the last value @xmath325 of a jacobi field with initial value @xmath305 to the initial speed @xmath326 , and can be deduced from algorithm [ alg : jacobi ] , which describes the function associating @xmath325 to the initial conditions @xmath327 and @xmath326 . to find the inverse of this function
, we consider the image of a basis of the tangent vector space in which @xmath326 lives .
+ + now , let us characterize the jacobi fields of @xmath5 to obtain the function @xmath24 .
a jacobi field is a vector field that describes the way geodesics spread out on a manifold .
consider @xmath328 , @xmath329 , a family of geodesics in @xmath5 , that is for each @xmath226 , @xmath216 \ni s \mapsto c(a , s,\cdot)$ ] is a geodesic of @xmath5 .
then for all @xmath59 , @xmath330 verifies the geodesic equations @xmath331 where @xmath332 is the srv representation of @xmath17 and @xmath219 is given by @xmath333 recall that we use the notation @xmath334 with @xmath335 for the tangential component of a tangent vector @xmath336 . to characterize the way these geodesics spread out , we consider the jacobi field @xmath302 \rightarrow t\mathcal m$ ] , @xmath337 by decomposing @xmath338 and @xmath339 we can write the second order variation of @xmath340 as @xmath341^{0,t } \label{jacobi } \\ & + \int_0^t \left [ \left.\big(\nabla_s\nabla_s\nabla_t j + \mathcal r(c_t , c_s)\nabla_sj + \nabla_s \left(\mathcal r(c_t , c_s)j\right ) \big)\right|_{a=0,t=\tau } \right ] ^{\tau , t } \mathrm d\tau . \end{split}\ ] ] the term @xmath342 can be expressed as a function of @xmath343 by twice differentiating the equation @xmath344 according to @xmath252 .
this gives @xmath345 since @xmath346 for @xmath227 , we know that the term we are looking for is @xmath347 . noticing that @xmath348 is equivalent to @xmath349 we get @xmath350 with @xmath351 .
\end{split}\ ] ] the terms @xmath352 and @xmath353 for all @xmath354 $ ] can be obtained by differentiating the geodesic equations and @xmath355 the first one gives @xmath356 and for all @xmath252 and @xmath100 we get @xmath357 the only term left to compute is the variation @xmath358 , which is by definition @xmath359 if we define @xmath360 for any fixed @xmath100 by @xmath361^{\tau , t } , \quad \tau\in [ t,1].\ ] ] since the covariant derivative of @xmath360 in @xmath362 vanishes , we can write for any @xmath363 @xmath364 integrating this equation according to @xmath362 from @xmath100 to @xmath365 we obtain @xmath366 where , since @xmath367 , we get for @xmath199 @xmath368 with finally @xmath369 we can notice that , however complicated , the numbered equations to when put together define a partial differential equation verified by the jacobi field @xmath340 .
they allow us to iteratively compute @xmath370 and @xmath371 , for a fixed step @xmath257 , knowing @xmath372 and @xmath373 . indeed
, we can estimate @xmath374 since @xmath375 is known for all @xmath100 , as well as @xmath376 since @xmath377 is known for all @xmath100 , and finally @xmath378 . assuming that we are able to compute the covariant derivative @xmath379 of the curvature tensor , for example if we are in a symmetric space ( then it is zero ) , we obtain an algorithm to compute the jacobi fields in the space of curves .
to summarize : [ alg : jacobi ] input : @xmath17 , @xmath380 and @xmath381 .
+ initialization : set @xmath382 and @xmath383 for @xmath102 $ ] .
+ heredity : for @xmath384 , set @xmath280 with @xmath281 and 1 . for all
@xmath100 , set @xmath385 \nabla_s\nabla_tj(s , t ) & = \nabla_t\nabla_sj(s , t ) + \mathcal r(c_s , c_t)j(s , t).\end{aligned}\ ] ] 2 .
compute @xmath386 for all @xmath102 $ ] .
3 . compute @xmath387 , @xmath388 and @xmath389 for all @xmath100 using , and , and deduce @xmath390 using equation .
4 . compute @xmath352 and @xmath391 for all @xmath100 using equations and .
compute @xmath392 using and @xmath393 , and deduce @xmath394 for all @xmath100 using equation .
finally , for all @xmath102 $ ] , set @xmath395^{s , s+\epsilon } , \\
\nabla_sj(s+\epsilon , t ) & = \left [ \,\,\nabla_sj(s , t ) + \epsilon \nabla_s\nabla_sj(s , t ) \,\,\right]^{s , s+\epsilon}.\end{aligned}\ ] ] output : @xmath325 .
using a discretization of algorithms [ alg : expmap ] , [ alg : geodshoot ] and [ alg : jacobi ] , we are able to compute an approximation of the optimal deformation between two curves , as shown in a toy example in figure [ fig : toyex ] . in this simple case
we perform geodesic shooting between two geodesics @xmath22 and @xmath23 ( in red ) of the hyperbolic half - plane @xmath6 . the reasons for our interest in this particular space , as well as the tools needed to work in it , are given in the next section .
the first two lines of figure [ fig : toyex ] show the different steps of the first iteration of geodesic shooting , and the last line gives only the last step of the following two iterations . in this simple case
, we can see that we converge in only three iterations .
further toy examples are given in figure [ fig : toyex2 ] , where we show the optimal deformations between pairs of geodesics of @xmath6 ( in blue ) , compared to the @xmath4-geodesics ( in green ) .
we can see in the first image that our metric has a tendency to `` shrink '' the curves in the center of the deformation compared to the @xmath4-metric .
we also show the influence of the orientation of the curves , on which the deformations depend .
we do not give the details of the discretization used for these examples or the ones in the following section here .
we refer the reader to @xcite for a detailed description of the discrete model .
in this section we consider the case where the base manifold @xmath0 is a symmetric manifold of negative curvature , the two - dimensional hyperbolic space @xmath6 .
we first explain why this space can be interesting for applications , namely as it coincides with the statistical manifold of gaussian densities equipped with the fisher information metric .
then we give some basic tools exponential map , logarithm map , curvature tensor needed to implement the previous algorithms in @xmath6 .
finally , we consider a specific application of curve analysis in that space , for the statistical study of locally stationary radar signals .
we explain how this framework gives us curves lying in the hyperbolic plane , and we present some simulation results .
it is possible to adopt a geometrical point of view to solve problems in various fields such as statistical inference , information theory or signal processing @xcite , @xcite , @xcite .
this framework is given by information geometry .
each element of a parametric family of probability densities @xmath396 can be seen as a point in the manifold of parameters @xmath397 . intuitively , it is easy to see that the euclidean metric is not always appropriate to compare two probability distributions in that space .
for example , each univariate gaussian distribution can be identified with its mean and standard deviation @xmath398 in the upper half - plane @xmath399 . as explained in @xcite , two univariate gaussian densities @xmath400 and @xmath401 with different means but the same standard deviation get `` closer '' to each other as their common standard deviation increases , meaning that intuitively the distance between the points of coordinates @xmath402 and @xmath403 in the upper half - plane should be greater than the distance between the points @xmath404 and @xmath405 for @xmath406 .
a more pertinent riemannian structure on the space of parameters @xmath397 is the one induced by the fisher information metric , defined in its matrix form as the fisher information . if the parameter @xmath407 is @xmath35-dimensional , @xmath408,\ ] ] for any @xmath409 and @xmath410 .
this metric is chosen , among other reasons , because it has statistical meaning : in parameter estimation , the fisher information measures the `` amount of information '' on the parameter @xmath411 contained in data sampled from the density @xmath412 ; it also gives a fundamental limit to the precision at which one can estimate this @xmath411 , in the form of the cramer - rao bound . in the case of univariate gaussian densities @xmath413 , fisher geometry amounts to hyperbolic geometry .
more precisely , the space of parameters @xmath398 equipped with the fisher information metric is in bijection with the hyperbolic half - plane via the change of variables @xmath414 .
indeed , with this rescaling of the mean , the fisher information matrix becomes @xmath415,\ ] ] which defines the riemannian metric of the well - known hyperbolic half - plane .
this is coherent with the example given above , since in the hyperbolic half - plane the distance between the points of coordinates @xmath416 and @xmath417 increases with @xmath418 for fixed values of @xmath419 , @xmath420 .
the differential geometry of gaussians has proved useful for applications , e.g. in image processing where in the image model , each pixel is represented by a univariate gaussian distribution @xcite , and in radar signal processing @xcite , @xcite , @xcite , @xcite , as we will see in section [ spectralestimation ] .
first , let us give a few tools which are necessary to work in the hyperbolic half - plane representation . along with the poincar disk , the klein model and others ,
the hyperbolic half - plane @xmath421 is one of the representations of two - dimensional hyperbolic geometry .
the riemannian metric is given by @xmath422 this means that the scalar product between two tangent vectors @xmath423 and @xmath424 at a point @xmath425 is @xmath426 using the usual formula ( see e.g. @xcite ) to compute the christoffel symbols , we can easily compute the covariant derivative of a vector field @xmath427 along a curve @xmath428 in @xmath6 .
it is given by @xmath429 where @xmath430 let us now remind a well - known expression @xcite for the riemann curvature tensor in a manifold of constant sectional curvature .
recall that @xmath6 has constant sectional curvature @xmath431 .
let @xmath432 be three vector fields on a manifold of constant sectional curvature @xmath433 .
the riemann curvature tensor can be written @xmath434 for the algorithms described above , we need to be able to compute the geodesic starting from a point @xmath435 at speed @xmath436 in other words , the exponential map @xmath437 as well as the geodesic linking two points @xmath176 and @xmath146 , with the associated initial vector speed the inverse @xmath438 of the exponential map .
the geodesics of the hyperbolic half - plane are vertical segments and half - circles whose origins are on the x - axis , as shown in figure [ fig : hyp ] , and they can be obtained as images of the vertical geodesic following the y - axis by a moebius transformation @xmath439 , with @xmath440 . to be complete ,
we give the proofs of the three following propositions in the appendix .
[ prop : geodh ] let @xmath441 and @xmath442 be two elements of @xmath6 .
* if @xmath443 , then the geodesic going from @xmath444 to @xmath445 is the segment @xmath446 with @xmath447 , and the logarithm map is given by @xmath448 * if @xmath449 , the geodesic is given by @xmath450 with @xmath451,\ ] ] where the coefficients of the moebius transformation can be deduced from the center @xmath452 and the radius @xmath63 of the semi - circle going through @xmath444 and @xmath445 : @xmath453 , @xmath454 , @xmath455 , @xmath456 , and for all @xmath102 $ ] , @xmath457 the logarithm map is in turn given by @xmath458 we now give the exponential map in @xmath6 .
[ prop : exph ] let @xmath441 be an element of @xmath6 and @xmath459 a tangent vector
. then the exponential map is given by @xmath460 , where * if @xmath461 , @xmath462 , * if @xmath463 , @xmath450 with @xmath464.\ ] ] the coefficients @xmath465 of the moebius transformation can be computed as previously from the center @xmath466 and the radius @xmath467 of the semi - circle of the geodesic , and for all @xmath102 $ ] , @xmath468 finally , we give the expression of parallel transport along a geodesic in the hyperbolic plane . [
prop : pth ] let @xmath469 be a curve in @xmath470 with coordinates @xmath471 , @xmath472 , and @xmath473 a tangent vector . the parallel transport of @xmath474 along @xmath151 from @xmath475 to @xmath100 is given by @xmath476 where @xmath477 . if @xmath151 is a vertical segment then @xmath478 , and if it is a portion of a circle , we get @xmath479 where the coefficients @xmath17 and @xmath35 of the moebius transformation can be computed as explained previously , and @xmath480 is the pre - image of @xmath151 by that transformation .
now that we have these explicit formulas at our disposal , we are able to test the algorithms described above in the simple case where the base manifold @xmath0 has constant sectional curvature @xmath431 .
note that computations are further simplified by the existence of a global chart . in radar signal processing ,
given an observation of a signal , it is useful to estimate the spectrum of the underlying process , as it is indicative of its structure
. if we are interested in the temporal modulations of that signal , we can estimate several spectra for that same signal and study their evolution in time .
the study of these time - frequency spectra , or spectrograms , is at the heart of micro - doppler analysis .
here we explain how a time series of spectra can be represented as a curve in the poincar polydisk .
the data we use for this example is synthetic data generated by a simulator of helicopter signatures . using this simulator
, we obtain a series @xmath481 of @xmath482 complex numbers that simulates the reflected signal received by a fixed radar antenna after sending a burst of @xmath482 pulses in the direction of a fixed helicopter .
given this vector of @xmath482 observations , the goal is to study the temporal evolutions of the underlying process .
to do so , we consider that this process is locally stationary and gaussian , and we estimate a spectrum for each stationary portion .
more precisely , using a gliding window of size @xmath483 to be adjusted , we estimate a high resolution spectrum for each position of the window of size @xmath484 on the vector of size @xmath482 .
this gives us a time series of @xmath485 spectra , which we index by @xmath486 .
let us now explain how we estimate and represent these spectra , each of which corresponds to the observation @xmath487 of a centered stationary gaussian time series @xmath488 .
to overcome the low resolution issues of the classical fft - based spectral estimation methods , burg suggested in the 1970s an alternative method based on autoregressive processes @xcite .
given the partial knowledge of the autocorrelation function of a stationary and gaussian process @xmath489 , burg showed that the process which maximizes the entropy that is , which adds the fewest assumptions on the data is an autoregressive process of the appropriate order .
following this maximum entropy approach , we estimate an autoregressive spectrum for each stationary portion @xmath490 using the so - called burg algorithm , see e.g. @xcite .
burg also showed that the second order statistics of such a process @xmath489 can be equivalently represented by its covariance matrix @xmath491 , a toeplitz ( because of the stationarity ) hermitian positive definite ( thpd ) matrix of size @xmath484 , or the so - called _ reflection coefficients _ @xmath492 of the autoregressive model , as there exists a bijection @xcite , @xcite @xmath493 , @xmath494 between the space @xmath495 of thpd matrices of size @xmath484 and the product space @xmath496 where these reflection coefficients live . here
@xmath497 is the unit disk of the complex plane .
this means that each stationary portion @xmath490 of size @xmath484 ( the size of the gliding window ) can be equivalently parameterized by its covariance matrix in @xmath495 or by an element of the product space @xmath496 .
we choose to work with the latter representation , because we can select a convenient metric on that space .
indeed , the legendre dual of the fisher information metric @xcite , defined on @xmath495 as the hessian of minus the entropy , @xmath498 has a nice expression in @xmath496 , in which the riemannian metric of the poincar disk appears @xcite @xmath499 with @xmath500 . in other words , equipped with the legendre dual of the fisher information metric
, the space of reflection coefficients becomes the product manifold @xmath501 , where @xmath502 is the poincar disk .
this means that each stationary portion @xmath488 of our radar signal can be parameterized in the product manifold @xmath501 by a set of coefficients @xmath503 , and so the entire locally stationary radar signal is represented by a time series in that space , which corresponds to a set of observations of the `` real '' evolution @xmath504 of the locally stationary signal .
+ with this choice of representation , comparing two vectors of radar observations can be carried out by computing the distance between the two corresponding curves in the product manifold @xmath505 , which , thanks to the product metric , is the same as comparing their components separately that is , pairwise comparing the evolutions of each reflection coefficient @xmath506 in the poincar disk . more generally , the representation of a vector of radar observation in a riemannian manifold enables to do basic statistics on these objects , such as defining the mean , median and variance of a set , or performing classification .
this can naturally be useful in target detection as well as target recognition .
here we use the algorithms presented in the previous sections to compute the frchet mean @xmath507 of @xmath176 curves @xmath508 .
the frchet mean , also called intrinsic mean , is defined by @xmath509 if @xmath5 is the space of curves in @xmath502 and @xmath35 the distance on @xmath5 .
since it is defined as a minimizer of a functional , this intrinsic mean can be found by a gradient descent type procedure , summarized as follows .
[ alg : karcher ] input : @xmath510 , @xmath511 . + initialize @xmath512 .
repeat until convergence : 1 . for @xmath513 , compute the geodesic @xmath514 linking @xmath512 to @xmath515 using geodesic shooting ( algorithm [ alg : geodshoot ] ) and its initial tangent vector @xmath516 .
2 . update the mean @xmath512 in the direction of the sum of the initial speed vectors @xmath517 using the exponential map ( algorithm [ alg : expmap ] ) .
output : @xmath507 . using our radar data , we compute the frchet mean of a set of @xmath518 curves tracing the evolutions of the reflection coefficient of @xmath518 signals we choose to represent only one of the coefficients @xmath519 , @xmath520 .
these signals are generated using the helicopter simulator , and correspond to the observation of 11 helicopters which differ only in their rotor rotation speeds .
we consider small variations ( less than @xmath521 ) around the mean value of @xmath522 rpm ( rotations per minute ) , and show the obtained curves in figure [ fig : karcherhelico ] .
theses curves are shown in the hyperbolic half - plane representation , which is equivalent to the poincar disk in terms of geometry .
the four images correspond to observations at different times of the @xmath518 helicopters for the same duration . in each case
, the red extremity of the colormap corresponds to the helicopter with the highest rotation speed , and the blue extremity to the lowest rotation speed .
the mean curve is shown in black in each image .
we have studied a first - order sobolev metric @xmath11 on the space of manifold - valued curves and its induced geometry .
the metric @xmath11 can be obtained as the pullback of a natural metric on the tangent bundle @xmath2 by the square root velocity function , and as such it is reparametrization invariant . the special role that @xmath11 gives to the starting points of the curves
induces a fiber bundle structure over the manifold @xmath0 seen as the set of starting points of the curves , for which the projection is a riemannian submersion .
the geodesic distance induced by @xmath11 takes into account the distance between the origins of the curve in @xmath0 and the @xmath4-distance between the @xmath523 representations , without parallel transporting the computations to a unique tangent plane as in @xcite and @xcite .
this should allow us to take into account a greater amount of information on the geometry of the manifold @xmath0 . using the pullback form of @xmath11
, explicit equations can be obtained for the geodesics , as well as for jacobi fields , which allow us to construct the optimal deformation between two curves by geodesic shooting .
once we can compute geodesics in the space of curves , we can also compute the mean of a set of curves and conceivably more .
these algorithms have been tested in the case where the base manifold @xmath0 is the hyperbolic half - plane , using toy examples as well as simulated radar data for the spectral analysis of locally stationary radar signals .
future work will include applications on the sphere for the statistical analysis of large trajectories .
this research was supported by thales air systems and the french mod dga .
the geodesic @xmath524 linking two points vertically aligned @xmath525 and @xmath526 is a vertical segment @xmath527 .
it verifies the geodesic equation @xmath528 . using the expression of the covariant derivative of a vector field in @xmath6
, this gives the equation @xmath529 , which can be rewritten as @xmath530 .
integrating twice , we find that @xmath447 . now if @xmath531 with @xmath532 , the geodesic @xmath151 is the image by a moebius transformation @xmath533 ( with @xmath534 ) of a vertical line @xmath535 , which gives @xmath536 we know that @xmath151 describes a semi - circle @xmath166 whose origin @xmath452 is on the x - axis , and that one end of the vertical line @xmath537 is sent to the point @xmath538 and the other to the point @xmath539 .
this implies that the center of the semi - circle is half - way between the two @xmath540 , and that the radius is @xmath541 .
these two equations as well as the condition @xmath534 gives a system of equations for the coefficients @xmath542 and @xmath35 , which , if we choose to set @xmath543 , yields the desired expressions .
if the extremity @xmath444 is sent by the inverse of the obtained moebius transformation on @xmath544 , and @xmath445 on @xmath545 , then the segment corresponding to the portion of @xmath151 linking @xmath444 to @xmath445 is @xmath546 .
taking the derivative of in @xmath547 gives the logarithm map .
the exponential map uses the same equations as the logarithm map with the difference that @xmath474 is known instead of @xmath445 .
the proof is very similar to the the proof of proposition [ prop : geodh ] and is not detailed here .
parallel transporting a vector @xmath473 along a curve @xmath151 from @xmath475 to @xmath100 is gives a vector field @xmath254 satisfying @xmath548 and @xmath549 . using equation ,
this can be rewritten @xmath550 where @xmath551 @xmath552 is of the form @xmath553 where @xmath554 is the identity matrix and @xmath555 and so the solution is @xmath556 , that is , @xmath557 with @xmath558 the matrix @xmath559 is diagonalizable and therefore its exponential can be written @xmath560 where @xmath561 and @xmath562 .
this gives us the desired formula .
99 s. i. amari , o. e. barndorff - nielsen , r. e. kass , s. l. lauritzen and c. r. rao , differential geometry in statistical inference .
institute of mathematical statistics , lecture notes- monograph series vol . 10
( 1987 ) .
j. angulo and s. velasco - forero , morphological processing of univariate gaussian distribution - valued images based on poincar upper - half plane representation , _ geometric theory of information _ , frank nielsen , springer international publ .
( 2014 ) 331 366 .
f. barbaresco , interactions between symmetric cone and information geometries : bruhat - tits and siegel spaces models for high resolution autoregressive doppler imagery , _ emerging trends in visual computing _ ,
lecture notes in computer science , * 5416 * ( 2009 ) , 124 163 .
m. bauer , m. bruveris , and p. w. michor , why use sobolev metrics on the space of curves , in _
riemannian computing in computer vision _
p. k. turaga and a. srivastava ) , springer - verlag ( 2016 ) , 233 255 .
a. srivastava , e. klassen , s. h. joshi , and i. h. jermyn , shape analysis of elastic curves in euclidean spaces , _ ieee transactions on pattern analysis and machine intelligence _ , * 33*,7 ( 2011 ) , 1415 1428 .
j. su , s. kurtek , e. klassen and a. srivastava , statistical analysis of trajectories on riemannian manifolds : bird migration , hurricane tracking and video surveillance , _ annals of applied statistics _ ,
* 8 * ( 2014 ) , 530 552 . | this paper focuses on the study of open curves in a riemannian manifold @xmath0 , and proposes a reparametrization invariant metric on the space of such paths .
we use the square root velocity function ( srvf ) introduced by srivastava et al . in @xcite to define a riemannian metric on the space of immersions @xmath1,m)$ ] by pullback of a natural metric on the tangent bundle @xmath2 .
this induces a first - order sobolev metric on @xmath3 and leads to a distance which takes into account the distance between the origins in @xmath0 and the @xmath4-distance between the srv representations of the curves .
the geodesic equations for this metric are given and exploited to define an exponential map on @xmath5 .
the optimal deformation of one curve into another can then be constructed using geodesic shooting , which requires to characterize the jacobi fields of @xmath5 .
the particular case of curves lying in the hyperbolic half - plane @xmath6 is considered as an example , in the setting of radar signal processing . |
the exciton dynamics is strongly affected by the spectral density and the dephasing rate in eq .
( 1 ) depends on the slope of the spectral density towards zero frequency .
[ fig : sfig1 ] shows a close up of spectral densities discussed in the main text , and in additions includes for convenient reference the single - peak drude - lorentz density ( dotted line , @xmath23 ) , used in some previous studies @xcite .
left panel : low frequency onset of the spectral density of the fmo complex .
right panel : global view .
circles : measured spectral density @xcite , parametrization @xcite .
solid line @xmath35 , dashed line @xmath39 , dotted line @xmath23 , parameters given in the main text .
, title="fig : " ] left panel : low frequency onset of the spectral density of the fmo complex .
right panel : global view .
circles : measured spectral density @xcite , parametrization @xcite .
solid line @xmath35 , dashed line @xmath39 , dotted line @xmath23 , parameters given in the main text . , title="fig : " ] the additional peaks present in @xmath78 at higher frequencies , which are not contained in @xmath72 , affect the coherences of the off - diagonal elements @xmath84 only weakly , fig . [ fig : sfig4zoom ] . the influence of the additional peaks in the spectral density @xmath78 becomes visible as small - amplitude oscillations for longer delay times ( fig .
[ fig : sfig4zoom ] ) .
this shows how the initially electronic coherence gets transformed into oscillations related to the details of the sharp peaks in the spectral density @xmath78 .
this finding is in - line with the theoretical calculations for spectral - densities with added @xmath81-peaked vibronic modes @xcite .
detail of the long - time evolution of the coherence @xmath70 , which shows small - amplitude oscillations for @xmath78 related to the peaks of the spectral density @xmath78 located at higher frequencies . for @xmath23 ( dotted line ) , almost no coherences prevail . ]
the same observation applies to the relaxation process , which is shown in fig .
[ fig : sfig2 ] for a population dynamics starting either at site @xmath85 or at site @xmath86 . for comparison , the results for the drude - lorentz spectral density ( dotted curve )
is also included , which deviates most significantly for @xmath87 .
population dynamics for an initial population at sites @xmath85 and @xmath86 for the @xmath88 spectral densities shown in fig .
[ fig : sfig1 ] and for temperatures @xmath2 k and @xmath3 k. the relaxation proceeds on very - similar time - scales for the two spectral densities @xmath35 and @xmath39 . , title="fig : " ] population dynamics for an initial population at sites @xmath85 and @xmath86 for the @xmath88 spectral densities shown in fig .
[ fig : sfig1 ] and for temperatures @xmath2 k and @xmath3 k. the relaxation proceeds on very - similar time - scales for the two spectral densities @xmath35 and @xmath39 . , title="fig : " ]
the cross - peak oscillations depend strongly on the initial slope of the spectral densities towards zero frequency .
this is reflected in fig .
[ fig : sfig3 ] , which compares the cross - peak oscillations for the two different spectral densities @xmath35 and @xmath23 , for which the dephasing- and relaxation contributions to the decoherence rate are reversed in magnitude .
[ fig : sfig3](b ) shows that for @xmath23 ( dotted lines ) the cross - peak oscillations are largely reduced at @xmath2 k and almost completely absent at @xmath3 k , while the super - ohmic onset of @xmath35 still leads to visible oscillations at @xmath3 k ( solid lines ) .
exciton dynamics for the fmo complex for two different spectral densities @xmath35 and @xmath23 .
( a ) real part of cross - peak cp(1,5 ) of the signal ( rephasing and non - rephasing ) at @xmath2 k and @xmath3 k. ( b ) coherent oscillations of the stimulated emission ( se ) part of cross - peak cp(1,5 ) . ]
moreover , also the amplitudes of the diagonal - peaks show pronounced differences for the two spectral densities .
the spectral density @xmath35 transfers the highest diagonal - peak amplitude upwards with increasing delay time from dp(2,2 ) to dp(3,3 ) as illustrated in the left panels of fig .
[ fig : sfig3 ] .
this shift is also seen in the experimental data at @xmath5 k @xcite , figure 1 , panels a , b .
+ in the drude - lorentz case ( fig .
[ fig : sfig3 ] , right panels ) the highest amplitude is located at diagonal - peak dp(2,2 ) for all delay - times , which corresponds to previous models ( see fig . 2 in @xcite ) , but differs from the experimental data and the result for @xmath35 .
( color online ) calculated 2d echo - spectrum of the fmo complex at @xmath3 k for the delay times @xmath89 fs .
shown is the real part of the rephasing and non - rephasing signal .
the square marks the diagonal peak ( dp ) with the highest amplitude .
left panels : spectral density @xmath39 , showing amplitude shifts from dp(2,2 ) at @xmath90 fs to dp(3,3 ) for @xmath91 fs , right panels : spectral density @xmath23 , highest amplitude remains at dp(2,2 ) . ] | the observed prevalence of oscillatory signals in the spectroscopy of biological light - harvesting complexes at ambient temperatures has led to a search for mechanisms supporting coherent transport through larger molecules in noisy environments .
we demonstrate a generic mechanism supporting long - lasting electronic coherence up to @xmath0 ps at a temperature of @xmath1 k. the mechanism relies on two properties of the spectral density : ( i ) a large dissipative coupling to a continuum of higher - frequency vibrations required for efficient transport and ( ii ) a small slope of the spectral density at zero frequency .
long - lasting oscillatory signals in optically excited light - harvesting complexes ( lhcs ) point towards the prevalence of a coherent electronic dynamics in molecular networks at physiological temperatures @xcite .
the experiments probe the electronic dynamics using two - dimensional ( 2d ) echo - spectroscopy for a series of delay - times between the excitation pulse and the probe pulse .
the 2d spectroscopy makes studies of the dynamics of dissipative systems possible and has found further applications in mesoscopic systems such as molecular nanotubes @xcite and semiconductor devices @xcite . due to
its known crystallographic structure and relative simplicity , the fenna - matthews - olson ( fmo ) complex serves as the prototype system for studying the choreography of the energy transfer from the antenna to the reaction center of a light harvesting complex @xcite . in 2d spectra long - lasting beatings
are observed , ranging from 1.2 ps at @xmath2 k to 0.3 ps at @xmath3 k @xcite .
the interplay of coherent dynamics , which leads to a delocalization of an initial excitation arriving at the fmo network from the antenna , and the coupling to a vibronic environment with slow and fast fluctuations , has lead to studies of environmentally assisted transport in lhcs @xcite .
an important open question is whether coherence plays a key - role in the functioning of light - harvesting complexes @xcite .
the theoretical understanding of the experiments is in its early stages and atomistic simulations based on molecular dynamics have not reached agreement @xcite .
a calculation of 2d echo - spectra based on the molecular dynamics simulation @xcite does not show clear coherent oscillations at @xmath3 k. one key ingredient for efficient transfer dynamics is the strong coupling to vibronic modes , which induces energy dissipation @xcite .
for the fmo complex the thermalization occurs within picoseconds and was observed by brixner et al . by the decay of diagonal - peak amplitudes to lower energies @xcite .
it has been proposed that the inclusion of the finite time scale of the reorganization process gives rise to long - lasting coherence in lhcs @xcite . while a sluggish bath relaxation leads to prolonged population beatings in the fmo network , calculations of 2d echo - spectra show oscillations of the exciton cross - peaks for only about @xmath4th of the experimentally recorded time at @xmath2 k @xcite . for an even longer bath - relaxation cross - peak oscillations
are absent already at @xmath5 k @xcite .
hence the search continues for theoretical models which support a long - lasting coherent dynamics leading to cross - peak oscillations in 2d echo - spectra and simultaneously retain the fast dissipation .
recently , the coupling to a superposition of discrete vibronic modes rather than the existence of a purely electronic coherence has been proposed as an explanation for the detected oscillations @xcite .
however , no significant changes in the oscillations of the fmo complex have been seen in experiments which altered the vibronic modes by using native mutants , @xmath6c substitution , and partial deuteration @xcite . here
, we present an alternative mechanism which does yield long - lasting and purely electronic coherence in the presence of a strong dissipative coupling required for efficient transfer . the mechanism is independent of the specific exciton - system and relies on properties of the continuum part of the vibronic spectral density as opposed to discrete vibronic modes .
realistic models of the exciton dynamics in the fmo complex have to include higher order phonon processes as well as the finite time scale of the reorganization process @xcite .
this requires to go beyond approximative rate equations @xcite and non - perturbative techniques @xcite are necessary to study the dissipative transfer dynamics . one key parameter determining the duration of coherent oscillations is the spectral density , which encodes the the mode - dependent exciton - bath coupling .
the spectral density has important implications for the two contributions to the decoherence rate @xmath7 which is the sum of the relaxation rate @xmath8 and the pure - dephasing rate @xmath9 ( @xcite , ch .
21.4.2 , @xcite ch .
3.8.2 ) , @xmath10 before we discuss decoherence - rates of the fmo complex , we analyze the impact of the spectral density on a two - site system with site - energies @xmath11 , inter - site couplings @xmath12 , and @xmath13 .
the two rates in eq .
( [ eq : dephasingrate ] ) are given by @xmath14 , @xmath15 , and are linked to the spectral density @xmath16 via @xmath17 . at zero frequency
, @xmath18 is proportional to the slope of the spectral density .
we evaluate the decoherence rate for two different models of the spectral density , ( i ) a super - ohmic spectral density @xmath19 , and a single - peak drude - lorentz spectral density @xmath20 , depicted in the supplementary material ( sm ) @xcite fig .
s[fig : fig1 ] . in the super - ohmic case @xmath21 and
the pure - dephasing term vanishes @xmath22 and decoherence arises only through relaxation .
the opposite regime is present for @xmath23 with @xmath24 . for the parameters used in models of a sluggish bath - relaxation in the fmo complex @xcite ( bath - correlation time @xmath25 fs , reorganization energy @xmath26
@xmath27 , @xmath28 @xmath27 , @xmath29 @xmath27 ) the pure - dephasing rate dominates @xmath30 @xmath27k@xmath31 , over the relaxation contribution @xmath32 @xmath27k@xmath31 to the decoherence rate . the spectral density for the fmo complex ( circles in fig . [ fig : fig1 ] )
has been extracted from fluorescence - line - narrowing experiments at @xmath33 k @xcite and shows a super - ohmic behaviour .
however , previous calculations of 2d - echo spectra of seven - site fmo complex have been often performed with a single - peak drude - lorentz spectral density @xcite , and did not yield long - lasting cross - peak oscillations .
the situation is different in the population dynamics , which has been analyzed for different spectral densities @xcite and yields no significant differences in the duration of population beatings for different spectral density .
spectral density of the fmo complex .
circles : measured spectral density @xcite , parametrization @xcite .
fits @xmath34 are used for the gpu - heom calculation eq .
( [ eq : specdens ] ) .
the marks on the frequency axis indicate the transition energies corresponding to the difference of exciton frequencies .
parameters for @xmath35 : @xmath36 @xmath27 , @xmath37 fs , @xmath38 @xmath27 , @xmath39 : @xmath40 @xmath27 , @xmath41 fs , @xmath42 @xmath27 . ]
as we show next , 2d echo - spectra are not directly derivable from population dynamics and carry along additional information about the exciton dynamics @xcite . to demonstrate this , we calculate the exciton - dynamics for the spectral densities @xmath39 ( @xmath35 ) shown in fig .
[ fig : fig1 ] , which represent the experimentally determined spectral density by a superposition of @xmath43 shifted drude - lorentz peaks @xcite @xmath44 the shifts allow us to vary the slope of the spectral density @xmath16 at @xmath45 and to mimic the super - ohmic behaviour seen in the experiment ( @xmath46 ) . compared to the single - peak drude - lorentz spectral - density discussed before ( @xmath26 @xmath27 , @xmath25 fs , @xmath47 ) , @xmath18 is reduced by a factor of @xmath48 @xmath49 . both superpositions
@xmath34 provide a good approximation of the spectral density at the transition energies corresponding to the exciton energy differences .
the hamiltonian of the fmo complex is based on the frenkel exciton model for lhcs @xcite and consists of a part describing the coherent exciton dynamics @xmath50 where @xmath51 is the creation operator of an electronic excitation of bacteriochlorophyll ( bchl ) @xmath52 with site energy @xmath53 .
the inter - site couplings between the @xmath54 bchls are denoted by @xmath55 .
the bchls are coupled to vibronic modes modeled by a set of independent harmonic oscillators @xmath56 .
the exciton - bath coupling is given by @xmath57 with @xmath58 .
the reorganization energy @xmath59 with @xmath60 is added to the site energy @xmath53 .
for the seven - site fmo complex we take the site energies and inter - site couplings from ref . @xcite .
for simplicity we assume identical vibronic - excitonic couplings at each site and neglect correlations of vibronic modes at different sites . the reorganization energy @xmath61
@xmath62 is of the same order as the excitonic energy - spacings , the bath - relaxation time , and the thermal energy @xmath63 at @xmath3 k @xcite .
the strong couplings of electronic and vibronic degrees require to solve the density - matrix propagation with a non - perturbative and non - markovian approach .
the computation of the 2d echo - signal for the fmo complex necessitates about @xmath64 more time - steps than the calculation of the population dynamics and has only recently been achieved for the non - markovian case @xcite .
we solve the exciton dynamics within the hierarchical equation of motion ( heom ) formalism @xcite .
we generalize this formalism from its original single - peak drude - lorentz term to a superposition of shifted peaks , eq . ( [ eq : specdens ] ) .
the implementation on high - performance graphics processing units ( gpu - heom ) @xcite allows us to advance 500 coupled hierarchical equations in lock - step and to obtain the computationally demanding 2d echo - spectra .
( color online ) calculated 2d echo - spectra of the fmo complex at @xmath2 k for delay time @xmath65 fs , spectral - density @xmath35 . ( a ) real part of the rephasing signal ( sum of ground - state bleach , stimulated emission , and excited state - absorption ) without static disorder ( linear scale ) .
the grid of exciton eigenenergies is labeled @xmath66 and cross - peak cp(1,5 ) marked by the arrow .
( b ) real part of the rephasing signal ( stimulated emission ) for static disorder fwhm@xmath67 @xmath62 .
the arcsinh scale facilitates comparison with @xcite , fig . 1 . ]
we verified the convergence of gpu - heom by increasing the hierarchy - depth and inclusion of low - temperature correction terms @xcite .
cross - peak cp(1,5 ) oscillations of the fmo complex , real part of the rephasing signal .
the sum of ground - state bleach , stimulated emission , and excited state - absorption is shown in ( a ) @xmath2 k and ( c ) @xmath3 k. ( b ) stimulated emission pathway at @xmath2 k for different disorder distributions fwhm@xmath68
@xmath62 .
( d ) stimulated emission pathway at @xmath3 k ( no disorder ) . ] a typical 2d fmo - spectrum at @xmath2 k and delay time @xmath65 fs for the spectral density @xmath69 is shown in fig .
[ fig : fig2](a ) .
the 2d echo - spectra are calculated as detailed in ref .
@xcite by evaluating the third - order response function , including excited - state absorption @xcite for the non - disorder averaged results .
we mimic rotational averaging by probing along eight molecular orientations with respect to the laser - field polarization .
the inclusion of static disorder leads to a broadening of the peaks as shown in fig .
[ fig : fig2](b ) for the stimulated - emission rephasing pathway and a gaussian disorder - ensemble of 200 realizations with fwhm@xmath67 @xmath62 added to the diagonal site - energies . the disordered peaks in fig .
2(b ) ( which does not include the excited - state absorption ) are somewhat narrower than in the experimental data @xcite , fig
. 1c . in fig .
[ fig : fig2 ] , we reference the peaks on the diagonal ( dp ) and the cross - peaks ( cp ) by their location on the exciton eigenenergy - grid .
disorder leads to a broad central region where the peaks coalesce ( compare fig .
[ fig : fig2](a ) and ( b ) ) , and amplitude is present in regions a , b , and c , in reasonable agreement the experimental data in @xcite , fig .
1c .
the calculated 2d - spectra differ significantly from the single - peak drude - lorentz results shown in the sm @xcite , fig .
s5 and in @xcite . besides the broader line - shapes in the single - peak drude - lorentz case
, the dynamics of the diagonal - peak amplitudes is altered and an upward amplitude transfer from dp(2,2 ) to dp(3,3 ) is visible for @xmath35 , which is in line with the experimental data at @xmath5 k @xcite , but does not exist in the single - peak drude - lorentz case .
exciton dynamics for the fmo with the spectral densities @xmath35 and @xmath39 .
time evolution of the coherence @xmath70 .
the duration of coherent oscillations of cross - peak cp(1,5 ) shown in fig .
[ fig : fig3 ] is in good agreement with the coherence life times @xmath71 . ] to determine the coherence - time of cross - peak oscillations , we analyze the changes in the amplitude of the cross - peak cp(1,5 ) marked with the arrow in fig .
[ fig : fig2](a ) . the oscillatory signal of cp(1,5 ) for @xmath72 shows modulations for the rephasing pathways ( fig .
[ fig : fig3](a ) ) .
the coherent nature of the oscillations is best analyzed in the stimulated - emission ( se ) pathway in fig .
[ fig : fig3](b ) which shows oscillations at @xmath2 k ( @xmath1 k ) for delay - times @xmath73 up to @xmath74 ps ( @xmath0 ps ) .
the excited - state absorption included in fig .
[ fig : fig3](a ) neglects correlation between excitons and almost compensates the stimulated - emission , resulting in a larger non - oscillatory background .
the addition of static disorder diminishes the oscillations to some extent , but the oscillatory signal prevails as shown for the se - pathways in fig .
[ fig : fig3](b ) at @xmath2 k. in the single - peak drude - lorentz case discussed in @xcite and the sm @xcite , fig .
s4 , the oscillations decay two times faster and no long - lasting oscillations remain in the stimulated - emission signal at @xmath3 k. the information of the exciton lifetime is already incorporated in the simpler to calculate single exciton dynamics . to this end , we prepare the system in an initial density matrix @xmath75 given in terms of two eigenstates @xmath76 of @xmath77 . in the 2d echo - spectra setup
such a delocalized exciton state is reached after the first two laser pules illuminate the sample .
the life time of the oscillatory components of cross - peak cp(1,5 ) of the 2d echo - spectra in fig .
[ fig : fig3 ] is in good agreement with the life time of the corresponding coherences @xmath70 shown in fig .
[ fig : fig4 ] . to isolate the impact of the vibronic mode - distribution we compare the coherences for the spectral densities @xmath35 and @xmath39 in fig .
[ fig : fig4 ]
. the coherences are almost unchanged by switching from @xmath72 to @xmath78 .
specific modes in the peaked spectral density as the one near @xmath79 @xmath27 induce beatings in the coherences with small amplitude @xmath80 .
in contrast to @xmath81-shaped vibronic modes , the peaks in @xmath78 which mimic the experimental data , have only a minor influence to the coherences .
however , this does not imply that higher - frequency modes are unimportant .
the relaxation to thermal equilibrium , which is prominently visible in 2d - echo spectra , relies on a large enough coupling of the exciton - system to the vibronic continuum .
the spectral densities studied have a reorganization energy @xmath82 @xmath27 which lead to a thermalization after @xmath83 ps at @xmath5 k as observed for the fmo complex @xcite . for efficient energy - transfer it is important to verify that the decoherence rate @xmath7 still supports a thermalization of the exciton system within a few picoseconds .
this is indeed the case for @xmath35 and @xmath39 and shows that despite the reduced slope of the spectral density , the system is still strongly coupled to the vibronic environment .
compared to the single - peak drude - lorentz form , the spectral densities @xmath34 support even a faster transfer from the initial bchl 6 to the target bchl 3 . in agreement with the experimental observations , the 2d spectra calculated here for the spectral density @xmath35 do support long - lasting oscillations of the _ cross - peaks_. in conclusion
, we have shown that two seemingly contradictory requirements , namely ( i ) strong coupling needed for a fast thermalization and ( ii ) the observation of long - lasting coherent oscillations are two sides of the same coin . the careful modeling of the continuous part of the spectral density towards zero frequency is essential for the prevalence of coherence in molecular networks since it determines the pure - dephasing rate @xmath9 . by combining the exciton hamiltonian with a spectral density with the small slope at @xmath45 seen in experiments , we find a much better agreement of the theoretically calculated 2d spectra and the experimental data once static disorder
is included .
the vanishing pure - dephasing rate of a super - ohmic spectral density implies that the decoherence is driven by the relaxation rate @xmath8 .
this is exactly the opposite behaviour compared to models based on the single - peak drude - lorentz spectral density @xcite .
furthermore , we find only a very small influence of the additional peaks in the spectral density @xmath39 on the coherences .
the cross - peak oscillations require a calculation of 2d - echo spectra beyond the weak - coupling expansion for structured spectral densities , which becomes feasible by the efficient gpu - heom algorithm .
we thank b. hein for help with calculating 2d - spectra and m. rodrguez for helpful discussions .
this work is supported by the dfg within the emmy - noether program ( kr 2889 ) .
time on the harvard seas `` resonance '' gpu cluster and support by the seas academic computing team are gratefully acknowledged .
10 see supplemental material in the appendix .
g. s. engel _
et al . _ ,
nature * 446 * , 782 ( 2007 ) .
h. lee , y .- c .
cheng , and g. r. fleming , science * 316 * , 1462 ( 2007 ) .
e. collini _
et al . _ ,
nature * 463 * , 644 ( 2010 ) .
t. r. calhoun _
et al . _ , j. phys .
chem .
b * 113 * , 16291 ( 2009 ) .
d. abramavicius _
et al . _ ,
phys .
rev .
lett . * 108 * , 067401 ( 2012 ) .
w. kuehn _
et al . _ , phys .
rev .
lett . * 107 * , 067401 ( 2011 ) .
g. d. scholes , g. r. fleming , a. olaya - castro , and r. van grondelle , nature chemistry * 3 * , 763 ( 2011 ) .
g. panitchayangkoon _
et al . _ ,
proc .
natl .
acad .
sci . * 107 * , 12766 ( 2010 ) .
p. rebentrost _
et al . _ , new j. phys . * 11 * , 033003 ( 2009 ) .
m. b. plenio and s. f. huelga , new . j. phys .
* 10 * , 113019 ( 2008 ) . c. olbrich , j. strmpfer , k. schulten , and u. kleinekathfer , j. phys .
chem .
lett .
* 2 * , 1771 ( 2011 ) .
s. shim , p. rebentrost , s. valleau , and a. aspuru - guzik , biophys .
j. * 102 * , 649 ( 2012 ) .
c. olbrich _
et al . _ , j. phys .
chem .
b * 115 * , 8609 ( 2011 ) .
c. kreisbeck , t. kramer , m. rodrguez , and b. hein , j. chem .
theory comput .
* 7 * , 2166 ( 2011 ) .
t. brixner _
et al . _ ,
nature * 434 * , 625 ( 2005 ) .
a. ishizaki and g. r. fleming , proc .
natl .
acad .
sci . * 106 * , 17255 ( 2009 ) .
b. hein , c. kreisbeck , t. kramer , and m. rodrguez , new j. phys .
* 14 * , 023018 ( 2012 ) .
l. chen , r. zheng , y. jing , and q. shi , j. chem .
phys . * 134 * , 194508 ( 2011 ) .
n. christensson , h. f. kauffmann , t. pullerits , and t. mancal , arxiv:1201.6325v1 ( 2012 ) .
a. chin _
et al . _ ,
arxiv:1203.0776v1 ( 2012 ) .
d. hayes _
et al . _ ,
faraday discuss .
* 150 * , 459 ( 2011 ) . c. w. gardiner and p. zoller , _ quantum noise _
( springer - verlag , berlin heidelberg , 2004 ) . y. tanimura and r. kubo , j. phys .
. jpn . * 58 * , 101 ( 1989 ) .
y. tanimura , j. phys .
. jpn . * 75 * , 082001 ( 2006 ) .
p. nalbach , d. braun , and m. thorwart , phys .
rev .
e * 84 * , 041926 ( 2011 ) .
l. mhlbacher and j. ankerhold , new j. phys .
* 10 * , 065023 ( 2008 ) .
w. koch , f. gromann , j. t. stockburger , and j. ankerhold , phys .
rev .
lett . * 100 * , 230402 ( 2008 ) .
j. roden , a. eisfeld , w. wolff , and w. t. strunz , phys .
rev .
lett . * 103 * , 058301 ( 2009 ) .
a. kolli , a. nazir , and a. olaya - castro , j. chem . phys . * 135 * , 154112 ( 2011 ) .
m. wendling _
et al .
_ , j. phys .
chem .
b * 104 * , 5825 ( 2000 ) . j. adolphs and t. renger , biophys .
j. * 91 * , 2778 ( 2006 ) .
u. weiss , _ quantum dissipative systems _ , 3rd ed .
( world scientific publishing co. pte . ltd . ,
singapore , 2008 ) .
v. may and o. khn , _ charge and energy transfer dynamics in molecular systems _
( wiley - vch , weinheim , 2004 ) .
j. yuen - zhou and a. aspuru - guzik , j. chem . phys . * 134 * , 134505 ( 2011 ) . c. meier and d. j. tannor , j. chem
. phys . * 111 * , 3365 ( 1999 ) .
s. mukamel , _ principles of nonlinear optical spectroscopy _
( oxford university press , oxford , 1999 ) .
b. brggemann , p. kjellberg , and t. pullerits , chem .
phys . lett .
* 444 * , 192 ( 2007 ) . |
solar flares are energetic events in the solar atmosphere .
energy release in flares leads to plasma heating , particle acceleration , and mass motions . during a solar flare ,
the plasma is heated and fills up the coronal loops .
the loops then evolve responding to this energy input and radiate emissions that can be observed .
the format of this released energy used to heat flares is still an issue of debate : do non - thermal particles carry a significant amount of energy and deposit most of it in the lower atmosphere and in turn heat the corona ? or does heating take place primarily in the corona and thermal conduction transfers heat flux into the lower atmosphere ? and in this process , how much energy is used to heat flare plasmas ?
these are important but unresolved problems .
many coronal heating models have been developed to study how released energy would heat flare plasmas resulting in observed radiation signatures .
modelers usually use an empirical heating function to describe the flare heating .
the heating function quantitatively depicts when and where a flare loop is heated , as well as for how long and by how much the loop heated .
it is a function of time and space . in time
, the heating rate is usually assumed as a single gaussian profile during the impulsive phase in numerical simulations ( e.g. , @xcite ) .
it has become recognized that heating may occur continuously throughout the life of a flare , starting from the pre - flare phase , and extending into the decay phase ( e.g. , @xcite ) .
@xcite found that the coronal temperature stays constant or even increases during the pre - flare phase .
they claimed that continuous heating in the corona is necessary to sustain the observed temperature .
in addition , @xcite proposed that the post - flare loop could be heated by magnetohydrodynamic shocks for a long time .
@xcite reported that continuous energy input is necessary to balance the rapid radiative cooling for a few flares .
continuous heating is also supported by spectroscopic velocity measurements .
@xcite found that the outer edge of the flare ribbon shows blueshifts in the decay phase , which indicate chromospheric evaporation responding to energy input .
these observations provide evidence of continuous heating throughout the lifetime of the flare . in space
, the heating function has a distribution along the flare loop .
furthermore , high - resolution imaging observations of the solar corona in the past two decades have led to important progress in our knowledge of flare plasmas : the flare plasmas are confined in numerous loops , whose basic scale is most likely below 1 @xcite .
these loops are usually formed and heated at different times and evolve independently of one another ( @xcite ) . in these different loops the heating rates are most likely different .
the concept of heating multiple loops has been applied in recent numerical studies ( e.g. * ? ? ?
* ; * ? ? ?
these models assume successive heating in newly formed loops to reproduce the observed coronal emission in the decay phase of the flare .
there are several ways to derive the empirical heating function of flares . in some previous studies , a heating rate linearly scaled , by a free parameter , to the observed
hard x - ray emission was used to model flare evolution ( e.g. * ? ? ?
in other studies that model multiple loop heating , the heating rates in these loops were prescribed by a number of free parameters that are adjusted to match model - produced coronal emission with observations ( e.g. * ? ? ?
these models are therefore constrained by comparing the model output the computed coronal loop emission , with observations .
recently , @xcite computed a heating rate using a reconnection model with measured reconnection flux in flares , and predicted the coronal radiation to compare with observations .
@xcite presented a new method to derive the flare heating rates using spatially resolved ultraviolet ( uv ) light curves , and then compute plasma evolution in the heated loops to compare with observations .
this recent method , compared with earlier ones , constrains the heating rates from both the input and output . in this paper
, we study heating of two sets of flare loops ( or two loop systems ) identified from imaging observations . each loop system is considered to be composed of a bundle of thin loops with their size given by the instrument pixel scale .
these loops connect opposite magnetic polarities .
we adopt the method of @xcite , with the heating rate observationally inferred from spatially resolved uv light curves at the footpoints of flare loops that are heated independently .
since the transition region responds rapidly to energy deposition and generates enhanced uv emission with a timescale much shorter than the observational cadence @xcite , the observed impulsive rise of the uv light curve indicates the starting time and duration of heating in each loop , and also roughly indicates the magnitude of heating namely , the brighter pixels are assumed to be heated with a greater heating rate . with this method
, the heating rates in numerous flare loops can be mostly inferred from uv observations and used to compute plasma evolution in these loops . in this study , there is only one free parameter to scale the peak heating rate to the observed peak uv emission , which is further determined by comparing model outputs with coronal observations @xcite .
therefore , parameters of the heating rates ( time , duration , and relative magnitude ) for individual loops are constrained from observations , and this data - driven flare heating function is temporally and spatially resolved .
we use these heating rates to compute plasma evolution in flare loops using the zero - dimensional ( 0d ) model called enthalpy - based thermal evolution of loops " ( ebtel ; @xcite ) in this study .
the ebtel model computes evolution of the average temperature and density in a coronal loop given an impulsive energy input , which has been validated by advanced one - dimensional ( 1d ) hydrodynamic simulations .
although the model works under very simplified assumptions , its high efficiency and very few free parameters allow us to calculate mean plasma properties for a large number of flare loops
. as these loops are heated at different times and evolve independently , at any given time , these different loops are in different evolution stages with different mean temperatures and densities .
therefore , this method naturally produces the differential emission measure as the temperature distribution of the plasma contained in the large number of loops , and allows us to calculate the total radiation in different wavelengths to compare with observations .
there is another reason to use this model : at present we do not clearly know how the heating rate distributes along the flare loop from observations .
therefore , the 0d ebtel model may be a sensible choice to compute plasma evolution to the first order .
furthermore , in this study , the number of available observables is much greater than the number of free parameters used in the model , hence comparison of the model with observed flare radiation and dynamic properties will provide a test of the model assumptions . through the model , the synthetic extreme - ultraviolet ( euv ) flux ( or intensity ) and mean enthalpy flow velocity can be computed and then compared with observations .
the presented observations here are high - resolution images from the atmospheric imaging assembly ( aia ; @xcite ) on board the _ solar dynamics observatory _ ( _ sdo _ ) and spectra from the euv imaging spectrometer ( eis ; @xcite ) on board _ hinode_. due to its observing mode , it takes some luck for eis to capture a flare . here , fortunately , eis observes a flaring region with a good temporal resolution .
it provides the information of flare radiation and dynamics as well as their evolution .
eis observations of this flare reveal some new results .
aia high tempo - spatial images provide more complete information on flare evolution and fine structure , which help constrain the model from both the input ( heating function ) and output ( coronal radiation ) .
such observational analysis combining with modeling provides better diagnostics of flare heating and dynamics .
the flare analyzed here is a c4.7 two - ribbon flare that occurred on 2011 february 13 .
figure [ fig_obs ] presents the observations .
the top panel plots the normalized light curves in _ goes _ soft x - ray 18 , aia uv 1600 and euv 171 wavelengths .
all three light curves exhibit two major peaks . in time sequence ,
the uv emission rises first , followed by the soft x - ray and then euv emissions .
this indicates heating followed by cooling of coronal plasmas .
the middle panel shows the images in the uv and euv bands from aia .
we can see the two ribbons and post - flare loops clearly .
the bottom panel gives the intensity and doppler velocity maps in the eis fe xii 192.39 line .
the velocity map reveals some upflows and downflows in the flare loops . for this event , although _ rhessi _ and _ fermi _ both show enhanced emission of up to 25 kev in the first few minutes as well as in the decay phase , there is no hard x - ray observation during most of the flare period because of _ rhessi _ night .
therefore , there is no strong evidence that thick - target non - thermal emission be significant during the flare .
aia observes the uv and euv images with the pixel scale of @xmath0 by @xmath0 and time cadences of 24 s and 12 s respectively
. such high tempo - spatial resolution allows us to identify the flare loops and their footpoints .
eis scanned over the flaring region using the 2 slit with a 4 step mode .
the spatial resolution is @xmath1 by 1@xmath2 in the directions across the slit and along the slit respectively .
it took eis 5 min and 27 s to scan over the region once , and the 19 scans covering the flare period are marked by the vertical dashed lines in figure [ fig_obs ] .
the eis data are corrected for dark current , hot pixels , and cosmic ray hits by using the standard reduction package .
the tilt of the eis slit is also corrected .
in addition , we correct the variations of spectral line positions throughout the orbit by averaging the line centers over a relatively quiet region ( the bottom 50 rows in eis raster ) . in this study , we select eight eis spectral lines , listed in table [ tab_line ] , to study the flare emission and dynamics .
all the line profiles , which exhibit a good symmetry , are fitted by a single gaussian function with constant background . when measuring the doppler velocity , we average the line center over the quiet region as the reference wavelength . here , only the unblended lines are chosen to determine the doppler velocity .
the co - alignment between aia and eis is achieved by comparing the 193 image and fe xii 195.12 intensity map .
lcc & @xmath3( ) & @xmath4(mk ) + fe x & 184.54 & 1.0 + fe xii & 192.39 & 1.25 + fe xiii & 202.04 & 1.5 + fe xv & 284.16 & 2.0 + fe xvi & 262.98 & 2.5 + ca xiv & 193.87 & 3.2 + ca xvii & 192.82 & 5.0 + fe xxiii & 263.76 & 12.5 + based on the flare morphology in euv images and the uv light curves at conjugate footpoints , we identify two flare loop systems .
their footpoints are marked by the two pairs of color boxes ( green and pink ) with a cross - sectional area of 4@xmath2 by 4@xmath2 ( limited by eis spatial resolution ) in figure [ fig_obs ] .
the loop segments are tracked by a series of black boxes ( each also with 4@xmath2 by 4@xmath2 ) .
hereafter we call the loop system located in the west ( with footpoints plotted in green ) as the first set of loops , and call the other one in the east ( with footpoints plotted in pink ) as the second set . these two sets of loops show very different behaviors in euv doppler velocity and also uv light curve .
figure [ fig_fp ] plots the averaged doppler velocities in multiple euv spectral lines at the four footpoint patches , together with the integrated light curves in uv 1600 .
it is seen that the north footpoints of the first set of loops exhibit blueshifts ( negative velocities ) of @xmath510 km s@xmath6 for about 25 minutes , followed by redshifts ( positive velocities ) of greater than 20 km s@xmath6 .
for the second set of loops , the north footpoints show stronger blueshifts of @xmath520 km s@xmath6 , which are maintained for about 1 hour .
note that , for both sets of loops , the intensity light curves are very similar at the northern and southern foot patches ; however , the measured doppler velocities at the south footpoints are significantly smaller than at the north footpoints , particularly in the second loop system .
this might be mainly due to the projection effect : first , different viewing angles of conjugate footpoints lead to different line - of - sight velocity components ; second , there might be ambiguity in the velocities measured at the south footpoints , since these measurements may include flows at the footpoints as well as part of the loops that appear to overlap the footpoints .
therefore , we only discuss the doppler velocities at the north footpoints of the two loop systems .
according to the chromospheric evaporation theory @xcite , the observed long - lasting blueshifts in the second set of loops suggest that continuous heating may be present for a long time .
this is also evident in the uv 1600 light curves at the footpoints of the loops .
the light curves of the second set of loops show two peaks before and after the flare onset ; while the light curves of the first set just brighten once during the rise of the soft x - ray flux .
the prolonged blueshifts observed at the footpoints of the second set of loops are coincident with the uv brightening .
the aia uv 1600 channel contains continuum and the c iv line . during a flare , the emission in this channel
is significantly enhanced , with the excess emission originating mainly from the upper chromosphere or transition region .
it is the most direct evidence of energy release .
the rapid rise of the uv emission reflects the energy deposition into the lower atmosphere . in this flare ,
the different evolutions of uv light curves of the two loop systems indicate different cases of energy release ( or heating ) : the energy release in the first set of loops occurs during the impulsive phase , consistent with the traditional point of view ; while in the second set of loops , the energy is continuously released during the flare , starting from the pre - flare phase , and extending to the decay phase .
this continuous energy release is also reflected by the doppler velocity measurements at the footpoints of the second set of loops , which show long - lasting blueshifts .
these observations suggest that the time history of energy release is different in these two sets of loops adjacent to each other .
the observations show that the heating process of this flare is very complicated , varying with different loops and spanning different phases of the flare
. measurements of the doppler shifts at the footpoints are averaged in the 4 by 4 area ( four eis pixels ) due to the coarse spatial resolution of eis ; however , these footpoint patches can be better resolved in aia images with a pixel scale of @xmath0 , which means that there are 49 aia pixels in a footpoint patch ( a green or pink box in figure [ fig_obs ] ) .
we consider that anchored to each aia pixel is a flare loop with the cross - sectional area of @xmath0 by @xmath0 ; therefore each loop system is composed of such 49 individual loops .
the spatially resolved uv light curves observed by aia can provide information of the heating history at the footpoints of these 49 loops .
we adopt the method of @xcite to observationally infer heating rates from uv light curves at the footpoints of these loops , and model plasma evolution with a 0d loop heating model , the ebtel model @xcite .
we then synthesize coronal radiation and enthalpy flow velocity to compare with euv observations .
the ebtel model describes the response of plasmas in the coronal loop to a given impulsive heating function .
the model solves an energy equation governed by the input heating rate and total loss rate .
the net loss includes the optically - thin radiative loss in the corona and an amount of loss through the transition region , part of which will be radiated by the chromosphere .
the loss through the transition region is either directly scaled to the coronal radiative loss with a free scaling parameter @xcite , or estimated by taking into account the atmosphere stratification @xcite . in this study , we have computed loop evolution with both versions of the code , and found that the difference is only about 10% in both the temperature and density values . in this paper , only the results computed using the second version of the model @xcite are presented .
the ebtel model also solves a mass equation by considering the flow across the boundary between the corona and its base , or the transition region ; this flow is computed in the form of the enthalpy flow as the difference between the conductive heat flux into the transition region and the loss from this region , when non - thermal particle precipitation is insignificant . in this study , lacking hard x - ray observations to identify the presence of non - thermal particles , we use the ebtel code to model flare loops primarily heated in the corona , and energy flux is transferred to the transition region primarily by conduction .
the model uses very few parameters ; it is highly efficient and particularly suitable for computing mean plasma properties in a large number of loops and deriving the sum radiation of these loops .
note that ebtel models a half loop assuming symmetric heating .
seen from figure [ fig_fp ] , conjugate footpoints of each set of flare loops exhibit rather similar uv light curves , hence we may justify the symmetry assumption , and model 98 half loops anchored at 98 aia pixels for each of the two loop systems .
the inputs of the ebtel model are half - loop length @xmath7 and volumetric heating rate @xmath8 .
they are derived from observations in our study . for this flare , the half lengths of these two sets of loops are both about 13 mm estimated from aia images with a semi - circle assumption .
the heating rate of a half loop is inferred from the uv light curve at the pixel where the loop is anchored .
it is expressed as @xmath9 , where @xmath10 is a shape function determined from the observed uv count rate at this pixel , @xmath3 is a scaling constant , which is the same for all 98 half loops in each set , that can be determined by best - matching modeled and observed total coronal radiation from the set of loops .
figure [ fig_hp ] shows the uv light curves from two of the half loops , each from a loop system .
all 98 loops in the first set exhibit a single peak at slightly different times during the impulsive phase , whereas about one half of the second set of loops exhibit two peaks as shown in the right panel of the figure , and the other half show only one peak in the uv light curve .
it suggests that the first set of loops are heated once , but twice - heating is present in one half of the second set of loops .
the @xmath10 function used here is different in these two sets of loops . in the first set
, it is assumed to be a full gaussian derived by fitting the rise phase of the footpoint uv 1600 light curve to a half gaussian .
note that , we do not use the decay phase to infer the heating rate for this set , since the decay phase is not a direct reflection of energy release but mostly a manifestation of plasma cooling .
while for the second set of loops , due to the complexity in the heating history , and more importantly , because we wish to examine whether significant continuous heating is present in the decay phase so as to produce long - lasting blueshifts , @xmath10 is assumed to be the observed uv light curve itself ( its effects on radiation will be discussed below ) . the finally derived heating rates ( red lines in figure [ fig_hp ] ) in these loops typically range from 0.3 to 1.3 @xmath11 . using the inputs described above in the ebtel model , we compute plasma evolution for each half loop .
as these loops are heated at different times and presumably by different amounts of energy , they evolve independently of one another , and exhibit different mean temperatures and densities at any given time .
the temperatures and densities of these 98 half loops are used to compute the differential emission measure ( dem ) of the coronal plasma for each set of loops with this formula : @xmath12 , where @xmath13 is the electron density , @xmath14 is the total volume of the plasma , and @xmath15 is the temperature ; in practical calculations , @xmath16 , where @xmath17 , @xmath18 is the loop cross - sectional area , @xmath19 indicates a loop that has a temperature falling in the range between @xmath20 and @xmath21 , and @xmath22 is the number of this subset of loops .
figure [ fig_ev ] shows the coronal dem for the two sets of loops averaged every 15 minutes during different evolution stages of the flare .
for the first set of loops , it is seen that the dem at high temperatures is not significant in the early stage ( 13:2013:35 ut ) .
however , it becomes significant with the heating added ( 13:3514:05 ut ) . in the late evolution stage ( after 14:05 ut ) , the dem decreases towards the low temperatures , which reflects cooling of plasmas .
for the second set of loops , the dem distributes in a wide temperature range in the early stage , and changes slowly at high temperatures during the subsequent evolution stages .
that is due to continuous heating in this set of loops .
from the computed temperatures and densities , we can derive synthetic coronal emission fluxes of the loops .
for each loop system , the fluxes from the 98 half loops are added up , and then compared with the euv emission observed by aia and eis .
the observed emission is measured from a series of 4 by 4 boxes along the loop system ( see figure [ fig_obs ] ) .
the model also computes the enthalpy flow velocity at the coronal base , which then can be compared with the doppler velocity measured by eis at the footpoints of the loop system .
firstly we compare the synthetic euv fluxes with aia observations .
figure [ fig_comaia ] shows the comparison in six euv channels .
note that there is saturation in some channels , e.g. , 211 , 193 , and 171 .
so we adopt the fluxes at short exposure times for these channels with smaller saturation effect . from the figure
it is seen that , for the first set of loops , the computed euv fluxes generally agree with the observed fluxes in the six aia channels .
they all exhibit similar evolution trends , as well as rise and reach peak at nearly the same time ; their magnitudes are also comparable .
there are discrepancies in a few places . in the 131 band which observes plasmas at high temperature ( @xmath510 mk ) ,
the modeled flux rises and decays somewhat faster than the observed . and
in the low - temperature bands ( 211 , 193 , and 171 ) , the model - computed flux rises a bit later than the observed , and exhibits a surge at around 14:10 ut , when the observed flux also peaks but with a smaller magnitude . for the second set
, it is shown that continuous heating in the model can produce double euv emission peaks as observed in all aia bands .
however , compared with the first set of loops , the model generates more flux in high temperature bands ( 131 and 94 ) , which are sensitive to plasmas at temperatures of 610 mk , and insufficient flux in the low temperature bands ( 211 , 193 , and 171 ) sensitive to temperatures of 12 mk .
in addition , for this set , the model - generated second peak is delayed with respect to the observed peak in a few bands .
we also compare the model results with eis observations .
figure [ fig_comeisi ] shows the observed and computed intensities of the eight eis lines listed in table [ tab_line ] for the two loop systems .
for the first set of loops , the computed intensity basically tracks the observed intensity in all eight lines ; in a few lines , the model under - estimates the intensity by within a factor of two . for the second set of loops , similar to the aia comparison ,
the model results and observations both show double peaks in a few relatively hot lines , though the second peak of the synthetic flux in these lines is somehow delayed by a few minutes compared with that observed .
similar to the aia comparison , the model generates very little flux in low - temperature fe x 184.54 , fe xii 192.39 , and fe xiii 202.04 lines . in summary , we see that aia and eis observations are rather consistent with each other , while there exist some discrepancies between model and observations
. the discrepancy may be partly caused by the mean temperature and density approach of the 0d ebtel model . in reality , there are temperature and density distributions along a single loop during its evolution , which may further broaden the dem curve and may improve the model - observation comparison .
it is also noted that , for the first set of loops , the heating rate is derived as a full gaussian by fitting only the rise phase of the uv light curve ; whereas for the second set of loops , the heating rate is directly scaled to the observed uv light curve from its rise to decay .
we have compared the model results for the second set of loops using different kinds of heating rates .
our experiments show that , these two different heating rate profiles have only minor effects on the radiation in high - temperature bands ; the heating rate directly scaled to the uv light curve reduces radiation in low - temperature bands , and delays the peak of low - temperature emission by a few minutes .
these effects on the low - temperature emission are caused by the extra heating into the decay phase , which leads to slower cooling of plasmas .
the computed radiation using this heating rate compares more favorably with the observed euv light curves than using the heating rate from half - gaussian fitting , suggesting that continuous heating into the decay phase may not be ignored in the second set of loops .
given the limitation of the model and the fact that only one free parameter , the scaling constant ( @xmath3 ) for the magnitude of heating rates , is used in modeling all the loops , we regard that there is rather good agreement between model results and observations , particularly for the first set of loops . besides euv emissions
, we also compare model - estimated flow velocity with observations .
note that , the eis doppler shift measurements are reliable in the four low - temperature lines , which are unblended and with high signal - to - noise ratios for this event ; while there are large uncertainties in the measurements for the other four high - temperature lines .
figure [ fig_comeisv ] shows the comparison of velocities in the four low - temperature lines .
the measured doppler shifts ( in color symbols ) are rather homogeneous for all the four lines without apparent temperature dependence .
the model computes the flow velocities at the coronal base of 49 half loops for each footpoint patch ; the solid dark line in the figure gives the average velocity of these 49 half loops . to estimate the velocity in the eis lines , we assume uniform pressure and steady - state flow across the coronal base , in which case , the velocity in each line is roughly proportional to the line formation temperature .
the colored dashed lines in the figure show the estimated velocities in the four lines at the footpoint patches for the two sets of loops . for the first set of loops
, it appears that the model - estimated velocities compare very well with the eis measurements , even though the assumption of steady - state equilibrium is rather inaccurate .
it is encouraging to note that the model - estimated and observed velocities both switch from upflow to downflow at around the same time .
the agreement is most prominent for the north footpoint where the projection effect is less significant .
for the second loop system , the synthetic velocities do not match well the measurements .
the synthetic flows are weak and short - lived , inconsistent with the observed strong and long - lasting blueshifts .
the bottom panel of the figure also shows flow velocities in all the 49 loops .
the flow velocities of individual loops are comparable with observations during the times of heating ; however , the synthetic flow speed does not match the observed phenomenon .
furthermore , observations show continuous upflows in between the two periods of heating , which can not be produced by the model . in summary ,
the model - observation comparison is different in the two sets of loops .
for the first set of loops , given the limitation of the 0d ebtel model and that only one free parameter is used in modeling nearly one hundred loops , the model results can be considered to be in very good agreement with the observed fluxes and flow signatures measured in multiple euv lines characteristic of plasma properties at different temperatures during different evolution stages .
these results suggest that the 0d multi - loop heating model is able to correctly describe the mean radiation and dynamic properties of this set of loops .
we note that ebtel only solves a gross energy equation without treating the detailed physics during the heating phase , and in this study , we model the loop heating with the assumption that heating primarily takes place in the corona with negligible precipitating non - thermal particles .
the agreement between model and observations in this set of loops would largely indicate that the model assumptions , such as heating in the corona and energy transport primarily by conduction , are not far from the real situation , and that in this case , the timing and total amount of heating energy of each of the heating events , which we infer from uv observations , are important for the net radiative output and dynamic evolution of these loops , regardless of the details of the heating phase .
for the second set of loops which are considered to be twice - heated , there are large discrepancies between model and observations particularly for the late heating phase and in low - temperature ( 1 - 2 mk ) lines : the model does not generate enough euv fluxes in these lines , and also fails to produce the long - lasting strong blueshifts measured in these lines ; the model - produced second emission peak is delayed by a few minutes with respect to observations .
these may indicate the complexity in plasma evolution with continuous heating not appropriately described by the present method .
we have investigated heating and evolution of two loop systems in a c4.7 two - ribbon flare through aia imaging and eis spectroscopic observations and ebtel modeling of multiple loops .
the two sets of loops show quite different behaviors : the first set exhibits 25-minute blueshifts at the footpoints followed by redshifts , while the second set shows blueshifts for about 1 hour indicative of continuous heating . in the mean time , the spatially resolved uv light curves from the feet of the two sets of loops show single and twice brightenings respectively , implying heating with different time scales . with heating functions inferred from the uv light curves at loop footpoints , we model the evolution of flare plasmas in these two sets of loops ,
compute the coronal radiation and enthalpy flow velocity , and compare them with observations .
the two loop systems show quite different evolution and dynamics in uv and euv observations , which demonstrate that there exist different heating processes in these two sets of loops .
the first set loops are mainly heated during the impulsive phase of the flare , while the second set loops appear to be heated twice , once during the pre - flare phase , and then again during the decay phase .
these are reflected in the footpoint uv emissions , and _ goes _ soft x - ray light curve also shows a few peaks corresponding to these different episodes of heating in these loops . as a consequence of different heating processes , the two sets of loops exhibit different dynamic behaviors .
spatially resolved observations allow us to further divide these different heating events into heating a few tens of loops , and therefore construct time and space ( loop ) dependent heating functions for the flare .
the evolution and dynamics of the first set of loops are verified by the model that produces emission and flow patterns consistent with observations .
the agreement suggests that the ebtel model , even though with the assumption of uniform heating and only computing mean properties in the loop , captures major features of the loop evolution in this set of loops , which are likely heated primarily in the corona and transfer energy downward by conduction .
however , the same modeling can not satisfactorily reproduce the observed euv flux and flow signatures in low temperature lines for the second set .
these may indicate different heating mechanisms in the two sets of loops that might not be treated by the same method .
the prominent discrepancies between model results and observations for the second set of loops may be caused by the following reasons .
( 1 ) the heating may be largely non - uniform along the loop , likely concentrated on the loop footpoints .
this is implied by the limited hard x - ray observations from _ rhessi _ , which show some bursty emissions up to 25 kev just before and after the flare impulsive phase .
based on this fact , non - thermal electrons may play some role in heating this set of loops .
however , such non - uniform heating can not be modeled by the present 0d ebtel model .
( 2 ) there may exist unresolved fine flaring strands , and the continuous heating takes place in different strands by very different rates as well as at quite different times .
the continuous heating in different unresolved strands would produce emissions at high temperatures by plasmas in newly heated strands as well as emissions at low temperatures by earlier heated strands that already cool down to 1 - 2 mk .
compared to the case of heating continuously occurring in the same resolved loop , this would produce long - lasting outflows as well as enough cool emissions , making the model results more consistent with observations .
the fine strands seems to be unresolved by the aia resolution @xcite .
we need higher resolution observations to infer the heating functions and model the plasma evolution in these fine strands .
though with simplified assumptions , the 0d ebtel model uses only one free parameter , and is very fast in computing mean plasma properties .
such high efficiency allows the model to be very useful for a quick - look of loop evolution , for parameter exploration and survey , and particularly for modeling a large number of loops @xcite .
it can constrain the flare heating model , and provide a reference for 1d or multi - dimensional models .
our study shows the advantage of using ebtel to model a large number of loops with reasonable agreement with observations in many aspects . on the other hand ,
the presented discrepancies between the 0d model and observations provide insights into how a 1d model may possibly address these discrepancies , improve model - observation comparison , and obtain more accurate diagnostics of flare heating and plasma evolution .
our spectroscopic analysis provides diagnostics of the flare dynamics .
plasma flow is crucially involved in the process of flare heating and evolution .
the velocity measurements , on one hand , can be used to determine free parameters in the model ( see @xcite ) ; on the other hand , they can help examine the assumptions made in the model , from which we deduce that the enthalpy flow velocity is proportional to temperature along the loop .
the linear relation between velocity and temperature along the loop has been tested to be valid in at least some observations ( e.g. , @xcite ) .
these signatures of temperature - dependent flow velocities can be better explored by 1d models using non - uniform heating function along the loop .
the potential of velocity diagnosis will be further utilized in the future .
the authors thank the referee for his / her constructive comments .
thanks james klimchuk for letting her use the new ebtel code , and dana longcope and wenjuan liu for their help with ebtel modeling and scientific discussions .
and m.d.d are supported by nsfc under grants 10878002 , 10933003 , 11373023 , and nkbrsf under grant 2011cb811402 .
is also supported by nsfc under grant 11103008 and csc under file no .
is supported by nsf grant atm-0748428 .
_ sdo _ is a mission of nasa s living with a star program .
_ hinode _ is a japanese mission developed and launched by isas / jaxa , collaborating with naoj as a domestic partner , and nasa ( usa ) and stfc ( uk ) as international partners .
scientific operation of the _ hinode _ mission is conducted by the _ hinode _
science team organized at isas / jaxa .
support for the post - launch operation is provided by jaxa and naoj ( japan ) , stfc ( uk ) , nasa , esa , and nsc ( norway ) . | we investigate heating and evolution of flare loops in a c4.7 two - ribbon flare on 2011 february 13 . from _
sdo_/aia imaging observations , we can identify two sets of loops .
_ hinode_/eis
spectroscopic observations reveal blueshifts at the feet of both sets of loops .
the evolution and dynamics of the two sets are quite different .
the first set of loops exhibits blueshifts for about 25 minutes followed by redshifts , while the second set shows stronger blueshifts , which are maintained for about one hour .
the uv 1600 observation by aia also shows that the feet of the second set of loops brighten twice .
these suggest that continuous heating may be present in the second set of loops .
we use spatially resolved uv light curves to infer heating rates in the few tens of individual loops comprising the two loop systems . with these heating rates , we then compute plasma evolution in these loops with the `` enthalpy - based thermal evolution of loops '' ( ebtel ) model .
the results show that , for the first set of loops , the synthetic euv light curves from the model compare favorably with the observed light curves in six aia channels and eight eis spectral lines , and the computed mean enthalpy flow velocities also agree with the doppler shift measurements by eis . for the second set of loops modeled with twice - heating ,
there are some discrepancies between modeled and observed euv light curves in low - temperature bands , and the model does not fully produce the prolonged blueshift signatures as observed .
we discuss possible causes for the discrepancies . |
rowd simulation has found its way into computer science , computer visualizations and the computer simulation of oriented building construction and crowd management @xcite . with continuously growing population around the world and with enormous evolution in the different modes of transportation in the last decade
a lot of paper have appeared with increasing interest in modelling crowd and evacuation dynamics .
thus the simulation of pedestrian flows has become an important research area .
pedestrian models are based on macroscopic or microscopic behaviour .
the evolution and design of any pedestrian simulation model requires a lot of information and data .
a number of variables and attributes arises from empirical data collection and need to be considered to develop and calibrate a ( microscopic ) pedestrian simulation model . for this reason we used different tools and developed different methods to collect the microscopic data and to analyse microscopic pedestrian flow .
it is very important to mention that the pedestrian data collection especially in a dangerous situation is still very much in its infancy .
an aim of this study is to establish more clearness and understanding about the microscopic pedestrian flow characteristics .
manual , semi manual and automatic image processing data collection systems were developed .
many published studies show that the microscopic speed obey a normal distribution with a mean of 1.38 m / second and a standard deviation of 0.37 m / second .
the acceleration distribution also resemblances a normal distribution with an average of 0.68 m/ square second @xcite . for the evolution and development of pedestrian microscopic simulation models , a lot of data
was collected with the help of video recording and tracking of moving entities in the pedestrian flow using the coordinates of the head path was established through image processing .
a large trajectory dataset has been restored . for the observation of pedestrian flows in public places
a sony camera was used .
this observation was in different places where the pilgrims perform their rituals .
many variables can be gathered to describe the behaviour of pedestrians from different points of view .
this paper describes how to obtain variables from video taking and simple image processing that can represent the movement of pedestrians ( pilgrims ) and its variables .
moreover in this work we try to understand several parameters influencing the pedestrian behaviour in riots or panic situations . for obtaining empirical data different methods
were used , automatic and manual methods .
we have analysed video recordings of the crowd movement in the tawaf in mosque / mecca during the hajj on the 27th of november , 2009 .
we have evaluated unique video recordings of a 105@xmath1 154 m large mataf area taken from the roof of the mosque , where upto 3 million muslims perform the tawaf and say rituals within 24 hours .
both microscopic video data collection and microscopic pedestrian simulation model generate a database called pedflow database . the properties and characteristics that are capable of explaining microscopic pedestrian flow
are illustrated .
a comparison between average instantaneous speed distributions describing the real world obtained from different methods , and how they can be used in the calibration and validation of the simulation tools , are explained .
typically , manual counting was performed by tally sheet or mechanical or electronic count board to collect density and speed data for pedestrian .
pedestrian behaviour studies are collected by manual observation or video recording in different public places like corridors side walks and cross walks .
the effectiveness of the data ( pedestrian speed ) collected on any observed area is strongly related to the number of pedestrians in the flow .
the relationship between speed , flow , and pedestrian density for a crowd population or human group has been published in many fundamental diagrams developed by fruin @xcite and others @xcite .
though for many reasons the method has been used to detect and count vehicles in automatic way can not be used to detect pedestrians , since this system has been evaluated through pneumatic tube or inductance loops .
as we can deduce from later work on this technology the possibility of applying this method to reproduce trajectory and motion prediction is still in a discussion phase .
other approaches use a neural network framework recursively to predict pedestrian motion and trajectory @xcite .
however the pedestrian trajectories in this system are calculated with incorrect simplifications .
in particular , only the nearest neighbour trajectories are considered .
the main shortcoming of such an estimation is that there is no uncertainty in this prediction , moreover a comparison of different path prediction shows this is still far from the reality in order to predict that all objects will follow the same set of paths exactly . a method which allowed people counting based on video texture synthesis and to reproduce motion in a novel way was introduced by heisele and woehler @xcite .
the method works under the assumption that people can be segmented from the moving background by means of appearance or motion properties .
the scene image is clustered based on the color and position ( r , g , b , x , y ) of pixel .
the appearance of each pixel in a video frame is modelled as a mixture of gaussian distributions .
a algorithm is used that matches a spherical crust template to the foreground regions of the depth map .
matching is done by a time delay neural network for object recognition and motion analysis .
a significant task in video intelligence systems is the extraction of information about a moving objects e.g. detecting a moving crowd with pedcount ( a pedestrian counter system using cctv ) was developed by tsuchikawa @xcite .
it extracts the object using the one line path in the image by background subtraction to make a space - time ( x - t ) binary image .
the direction of each travelling pedestrian is realized by the attitude of pedestrian region in the x - t image .
they reported the need of background image reconstruction due to image illumination change . an algorithm to distinguish moving object from illumination change is explained based on the variance of the pixel value and frame difference .
the electronic and digital revolution in video techniques during recent years has made it possible to gather detailed data concerning pedestrian behaviour , both in experiments and in real life situations @xcite .
the big challenge is to develop a new efficient method of defining and measuring basic quantities like density , flow and speed .
basic quantities of pedestrian dynamics are the density @xmath2 [ 1/m@xmath3 in an area @xmath4 and the velocity @xmath5 [ m / s ] of persons or a group of persons , and the flow through a door or across a specific line @xmath6 [ 1/s ]
. the measurements also yield mean values of these quantities .
the task is to improve the given methods such that they allow to go fairly close to the real data of the crowd quantities .
the methods presented here are based on video tracking of the head from above .
note that tracking of e.g. a shoulder or the chest might be even better , though more difficult to obtain .
the density distribution knowledge in a very crowded area allows us to draw a so called density map to show us congestion directly as regions of high density .
the relationship between the pedestrian density @xmath2 and the pedestrian maximum walking speed @xmath7 are formalized into a graph known as the fundamental diagram @xmath8 @xcite .
since pedestrians move slower in a region of high density , the simulated particles should update their speed with the surrounding circumstances to maximize their rate of progress towards their goals .
tawaf observations at the haram mosque in mecca were made during hajj 2009 by mr .
faruk oksay .
the mataf area has 10 entrances / exits .
the flow of the tawaf is controlled .
all pilgrims begin and end their tawaf at the same place ( see fig.[fig : haramentrances-1 ] ) .
the number of pilgrims during this period is sufficient to observe the behaviour of high density crowd dynamics .
figure [ fig : haramentrances-1 ] . shows the main gate doors , side entrances , stairs to the mataf open air of the haram .
all observations took place on friday november 27th 2009 corresponding to 10th of dhu al - hijjah
1430 hijri in the afternoon . during the total observation period of three hours , three prayers ( midday , asr and maghreb ( sunset - prayer ) ) were performed , where in this time the mataf area comes to a standstill ( see fig . [
fig : prayer-4 ] ) .
our video observations show that the pilgrims have the desire to be near the kaaba .
therefore approximately 70 percent ( visually detected on video ) of the pilgrims perform their tawaf movement near the kaaba wall , which causes a high density in this area . in figure
[ fig : prayer-4 ] , one can see all of the pilgrims perform the prayer ritual in the holy mosque in mecca .
the tawaf around the kaaba is a periodic movement for the time between two prayers .
the observed number of pilgrims performing their tawaf ritual at the mataf area increases slowly after every prayer until the mataf attains it s maximum capacity ( see fig .
[ fig : mataf3 ] ) .
figure [ fig : mataf3 ] shows a typical pedestrian movement in the mataf area over daytime . during prayer times individuals stand still and therefore movement equals approximately zero .
the fluctuations in the velocity flow are created by the turbulence in the pedestrian flux .
note that the average local density in a specific location in the mataf area exceeded 8 persons / m@xmath9 during the hajj periods ( see fig .
[ fig : densitydistribution-5 ] and [ fig : density-1 ] ) . our first goal is to identify new methods and create a test system capable of extracting pedestrian movement information from video , similar to that collected by our hd - cameras in the hajj-2009 , such that any movement can be analysed to spot suspicious activity .
this task to collect pedestrian data and extract pedestrian motion from video sequences required an involvement and development of appropriate methods , followed by further analysis of this data to identify emergent motion or crossing trajectories .
the secondary goal is to identify the limitations of the approach including the system and data requirements for the techniques to work more effectively .
more specific , the project goals are : * develop a framework for video and image analysis , * develop an approach and relevant diagnostic software to collect movement data from video , * identify the requirements for such methods to work effectively , such as image quality , resolution and orientation , * identify how to interpret movement information , * interpret the movement data and examine abnormal behaviour , * design and produce a working implementation that demonstrates the above goals , * identify approaches that could further improve the system .
there are different techniques developed to extract information describing the position of pedestrians in a location , but not all of them are appropriate for detecting and pursuing pedestrian movement under different and extremely weather conditions . in their published work
@xcite , papageurgiou and poggio developed a system attempting to recognize human figures based on pixel similarities through a large training set of figures under various light and weather conditions . to identify the movement of the figures , the system analyses the similarity between matches of consecutive frames .
this method works quite well when the training set is large , but requires a high computational efficiency which achieves processing rates of 10 hz @xcite .
the study shows that accurate recognition can be done with coarse image data .
another approach to estimate crowd density is based on texture analysis .
velastin et al .
@xcite assumed that crowds with high density possess texture properties .
the proposed method , texture features were computed for the whole image and applied to crowd density estimation @xcite . in particular , all displayed textures , like wavelets @xcite and the gray level dependence matrix @xcite , were used to estimate crowd density .
the results exhibit , how effective statistical analysis of texture display is compared to neural networks when measuring crowd density .
unfortunately , this system examines only static images and can not cover crowd motion , but the techniques can be used to track pedestrian movements .
other strategies based on image segmentation were pursued by heisele and woehler @xcite , where raw data is filtered to split the image into segments , which are then analysed .
those images that match particular shapes are analysed further .
this approach allows to distinguish different images with common color and luminescence .
the required data on pedestrian behaviour ( e.g. density - effect , shock - waves - effect , ... ) in the haram can be done from our video recordings .
all observed effects can be analysed by simply watching the recorded videos .
but if we want to extract data like walking speeds from such observations we have to examine the videos frame by frame .
this is very time consuming . as a result of this , and
the need for more efficient data , the idea arose to use an automatic detection system . at that time
no sufficient system was available for the detection of human bodies , therefore some essential requirements were formulated . from the requirements we derived an idea to formulate an image processing system with the help of other programs , such as optical flow with opencv ( http://opencv.org/ ) and quest3d ( http://www.quest3d.com/ ) .
the materials used for this test are videos recorded at an outdoor piazza of the haram mosque in mecca where people congregated at different times during one day , simulating a surveillance application .
the data content had a wide range of crowd densities , from very low to very high .
three different data - sets , labelled morning observation , afternoon observation and combined observation ( before and after the prayer times ) were used .
each data set had 20 selected images with high resolution .
examples of images are shown in figure [ fig : muster-1 ] and [ fig : grid-1 ] . in order to collect pedestrian data and to study pedestrian traffic flow operations on a platform in detail , observations were also made from a platform of the haram mosque in mecca . these observations concerned pilgrim walking speeds and density distributions on the mataf area and ( individual ) walking times as functions of the distance from the kaaba wall .
the estimation of crowd density is an important criterion for the validation of our simulation tools .
processing is done in three levels .
* existing footage is loaded on a 3d program as a backplate . * from several provided 2d- architectural drawings we build a 3d model of the mosque . * a virtual camera has to be matched in position , rotation and focal length to the original camera so that the features of the 3d - model match the features of the filmed mosque . as the dimensions of the mosque are known , we then establish a grid of regular cells on the mataf area , each one of which has a size of 5mx5 m ( see fig . [
fig : raster-1 ] ) . through image editing software , we start a manual counting process . this regular grid is used to observe the density behaviour over all of the mataf area , from the nearest range to the kaaba wall up to outside of the mataf and the accumulation process ( by the black stone and maquam ibrahim ) .
the results of this investigation are shown in figures [ fig : densitydistribution-5 ] ( a ) , ( b ) , ( c ) and ( d ) and illustrate us the behaviour of the pilgrim density on the mataf area at different times during the day .
5 m . ] with a new computer algorithm developed within this investigation , where the mataf area is divided in regular cells . the number of pedestrians in every cell as function of time is determined through repeating the counting process many times .
the average value is identified as local density @xmath10 .
the data extracted from the videos allowed us to determine not only densities in larger areas , but also local densities , speeds and flows .
as an example the density distribution on the mataf area is shown in figure : [ fig : densitydistribution-5 ] .
the data was obtained by semi - manual evaluation .
[ [ dependence - of - the - density - distribution - on - the - mataf - as - function - of - time ] ] dependence of the density distribution on the mataf as function of time + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + figure [ fig : densitydistribution-5 ] shows density decline curves for different distances from the kabaa in a specific time .
the curves indicate that the local density amount vary strongly over the ( 0 < x < 40 m ) range .
figure:[fig : density-1 ] shows the pedestrian density distribution on the mataf area as a function of the position @xmath11 and time @xmath12 .
one clearly recognizes density waves , with maximum density package near the kaaba wall .
there the average local density can reach a critical value of 7 to 8 persons / m@xmath9 .
the congested area increases the local density to a critical and dangerous amount . as a consequence
the pedestrians begin to push to increase their personal space and create shock - waves propagating through the crowd , which can be seen as density waves , or density packages .
the density map illustrates how the pedestrian density decreases from the inside to outside of the mataf area , ( see fig.[fig : densitymap-1 ] ) . as we have mentioned that in the mataf area pedestrians move in the restricted space , the layout is gradually painted in different colors .
the color of every point of the space corresponds to the current density in this particular area .
the density map is constantly repainted according to the actual values : when the density changes in some point , the color changes dynamically to reflect this change . in case of zero density the area is not painted at all ( see fig . [
fig : densitymap-1 ] ( a ) , ( b ) , ( c ) and ( d ) )
. during the rush hour in a hajj period the local density in the mataf area reaches the maximum as we can see in the following figures [ fig : density-1 ] ( d ) and [ fig : densitymap-1 ] ( d ) .
the local density can reach 8 to 9 persons / m@xmath9 in a specific time during the day .
the maximal density concentrates near the kaaba wall .
[ [ densities - over - time - and - space ] ] densities over time and space + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we observe the density behaviour on the mataf area at different times during the day , before and after the prayer , and we compare this density with the simulation density results .
the maximum registered density was 7 to 8 persons / m@xmath9 and this represents a high crowd density .
the results of the estimation based on the statistical method , presented in figures [ fig : densitydistribution-5],[fig : density-1 ] and [ fig : densitymap-1 ] , reached a mean of 92 percent correct estimations .
it is possible to verify that the results were quite good for all evaluated images except for the one made up of high density crowd images , which reached only 84 percent correct estimations . in the mataf area , near the black stone , the pilgrim density reached over 9 persons / m@xmath9 .
for this reason it is very difficult to recognize and track every head and as a result , a 100 percent correct estimation would be very difficult .
all statistical results illustrating the density distribution at the mataf area at different time intervals are demonstrated in the figure [ fig : densitydistribution-6 ] .
this part of the dissertation considers the role of automatic estimations of crowd density and their importance for the automatic monitoring of areas where crowds are expected to be present .
a new technique is proposed which is able to estimate densities ranging from very low to very high concentrations of people .
this technique is based on the differences of texture muster on the images of crowds .
images of low density crowds exhibits rough textures , while images with high densities tend to present finer textures .
the image pixels are classified in different texture classes , and statistics of such classes are used to estimate the number of people . the texture classification and the crowd density estimation
are based on self - organizing neural networks .
results obtained estimating the number of people in a specific area of the haram mosque in mecca are presented in figure [ fig : colordensitydistribution-1 ] ) . in the latter paragraphs we focus on crowd density estimation for several reasons .
according to the crowd disasters study by helbing and johansson @xcite , one of the most important aspects to keep a crowd safe is to predict and identify areas with high density crowds preventing large crowd pressures to be built up .
areas where crowds are likely to build up should be identified prior to the event or operation of the venue .
this is important as crowds usually exist in certain areas or at particular times of the day .
places where crowd density rises up over time are likely to congest and need careful observations to ensure the crowd safety .
basically , crowd density surveillance and estimation can be a good solution for management and controlling the crowds safety .
the results of the estimations obtained during the tests allow us to consider both methods successfully .
while the statistical method reached quite good estimation rates ( around 92 percent ) for most groups , the spectral method illustrated small deviations between the best and the worst estimations , reaching on average almost the same rates of correct estimation obtained by the statistical method .
as speeds are hard to observe , walking times were measured , from which walking speeds were derived .
in addition to walking times and pedestrian densities other variables needed to be considered to complete the input of the simulation model ( such as the number of in and out going pilgrims and the configuration of the structure during the rush hour at the hajj period ) .
the observables are the walking time , velocities and the corresponding densities of the pilgrims performing their tawaf and say .
the movements of the pilgrims going in and out of the haram give us data to calculate the flux related to the tawaf .
the distribution of both in and out going pilgrims over the haram can be derived from this data . the second type of observation concerns individual walking times . in order to measure the pilgrims walking times in and out of the haram , pilgrims were recorded from the moment they started walking from one spot to another , either on the piazza or going up the stairs .
the start and duration of activities , such as tawaf or say , were measured also .
finally , locations of origin , destination and possible activities of the pilgrims were registered .
to do this , the piazza is divided into small areas with a length of 5@xmath15 meters .
we also recorded the movements of the pilgrims at specific moments , such as prayer times when the number of pedestrians increases dramatically .
therefore , cumulative flow curves can be constructed , out of which densities can be derived .
these curves can be compared with the reference curves of predtetschenski - milinski @xcite .
data was collected on a specific subject group of pedestrians who appeared to be 40 years of age or older . on the roof of the mataf area we selected our tracking subjects , consisting of adult men , women and people in wheelchairs .
the following individuals were specifically not considered : * children under 13 years of age , * pedestrians carrying children , heavy bags , or suitcases , * pedestrians holding hands or assisting others across the mataf , * pedestrians using a quad pod cane , walker , two canes , or crutches . to accurately quantify the normal walking speeds of the various subject groups , pedestrians who exhibited any of the following behaviour
were also not considered : * crossing of the mataf path diagonally , * stopping or resting in the mataf area , * entering the roadway running ( anything faster than a fast walk ) , the pedestrian sex ( male or female ) of each individual in the mataf area was recorded , as well as whether he or she was walking alone or in a group .
the group size was also noted when applicable .
a group was defined by two or more pilgrims walking the mataf trajectory at about the same time , regardless of whether or not they were apparently friends or associates . in the mataf area
, the pedestrian groups can reach 30 pilgrims walking together in the pedestrian stream .
in addition , subjects paths were monitored to determine when they started and ended their tawaf .
being inside the mataf was defined as being within or on the painted tawaf walking lines .
other pedestrian behaviour was recorded when if occurred : * confusion ( hesitation , sudden change in direction of travel or change of point of interest ) exhibited before walking , * confusion exhibited after entering the mataf trajectory , * cane use , * following the lead of other pedestrians , * stopping in the walking path during the tawaf movement , * difficulty going into mataf , * difficulty going out of the mataf .
several methods were developed to check the accuracy and performance of walking speed estimation abilities of the observers .
first , the walking speed was measured at the same time by three observers , then correlations between the estimates of all observers were determined . in particular , the walking velocity of one pilgrim was measured by the three observers and the mean value was taken .
the results of these verification procedures are discussed after the next section . from our video recordings
we choose places between two minarets as references , ( see fig .
[ fig : minaret-1 ] ) .
as the dimensions of the mosque were known , we then established a grid of regular cells covering all of the mataf area , each one having a size of 5mx5 m ( see fig .
[ fig : grid-1 ] ) .
the distance between the two minarets is known .
pedestrian crossing times were measured with a digital timer and an electronic stopwatch was implemented and synchronized with the timer of the video recorder .
the watch was started as the subject stepped off the first minaret and stopped when the subject stepped out on the opposite minaret after crossing all the distance between the two minarets .
are measured .
since the distance between the two minarets is known from the architectural plan of the haram , the average of the local pedestrian velocities @xmath13 can be determined . ] from the roof of the mosque every pedestrian can be identified . to establish the ability of the field observers to identify the fitness level or the age of pedestrians with high accuracy
a simple verification procedure was performed .
the age estimation and the level of fitness of the pedestrians was based on their walking speed .
it is a physio - medical fact that older pedestrians walk more slowly than younger ones ( this is easily supported by field data ) , however , the published or already existing data on walking speeds and start - up times ( i.e. the time from the beginning of a tawaf movement until the pedestrian steps off the mataf ) have many shortcomings .
here we consider the complicated movement of the tawaf and the human error rate of the observer .
the walking speed on the mataf area can be affected by many factors , one of the relevant factors is the age of the pedestrian .
this demonstrates that the observations were quite good at identifying older pedestrians or pedestrians with fitness deficiency or physical health problems .
a digital stopwatch was integrated with the video recording sophisticated for the measurements of pedestrian crossing times .
the crossing times of the same pilgrims were measured during five rounds of the tawaf and the average value was determined . 5 m . with the help of the regular cells and the distance between two minarets in the haram ( see fig .
[ fig : minaret-1 ] ) the ( individual ) walking times @xmath14 are determined and the average of the local speeds @xmath13 is calculated .
the average walking speed for male pedestrians is 1.37 m / s , female 1.22 m / s and for people moving on wheelchairs 1.534 m / s . ]
this research also examined the impact of the building layout on the pedestrian speed distribution and the pedestrian density of pilgrims performing the tawaf movement around the kaaba .
the set of data of pedestrian walking speeds which were obtained through analysing video recording using a set of statistical techniques are displayed in figures [ fig : velocitydismen-1 ] ( a ) , ( b ) and ( c ) .
the results revealed that walking speed seems to be following a normal distribution no matter of male , female , older or younger .
the average speed of young people is dramatically larger than that of older people , and the average speed of male is slightly larger than that of female .
the width of the obtained curves is related to the different standard deviations .
the mean computed walking speed represents the speed that 85 percent of pedestrians did exceed .
a total of 250 pedestrians were observed .
included were 100 male pedestrians of about 60 years of age , 100 women pedestrians and 50 wheelchair pedestrians .
this data describes all of the pedestrians observed : those walking in the center of the stream and those walking by the edge of the mataf trajectory .
as is subsequently described , those who were walking by the edge of the mataf tended to walk more quickly . all observed pedestrians moved in a rotational motion around the kaaba counter - clockwise ( tawaf ) , in compliance with the pilgrim stream .
the mean walking speed for male pedestrians was 1.37 m / s and 1.22 m / s for female pedestrians . in conjunction with pilgrims old , the mean walking speed for younger pedestrians was 1.48 m / s and 1.20 m / s for older male pedestrians .
the results revealed that the average walking speed for young women are 1.32 m / s and 1.12 m / s for old women .
this means * young male pedestrians had the fastest mean walking speeds [ 1.48 m / s ] and older females had the slowest [ 1.12 m / s ] . the differences between young men and young women [ 0.16
m / s ] and between older men and older women [ 0.1 m / s ] , this result shows a little deviation that can be traced back to the fitness level of pedestrian or other factors , in the normal condition are approximately the same .
the mean walking speed for the younger pedestrians ranged from 1.37 to 1.57 m / s across all conditions , with an overall mean speed of 1.48 m / s .
the means for the older pedestrians range from 0.97 m / s to 1.26 m / s , with an overall mean speed of 1.18 m / s . for design purposes a mean speed of 1.33 m
/ s appeared appropriate ; * locations by the edge of the mataf had faster walking speeds because such locations has a lower pedestrian density .
it is clear that the pedestrians near the kaaba had a short walk path but in this places densities of 7 to 8 persons/ m@xmath9 can be exceeded , making the movement of pilgrims very slow and turbulent ; * places situated further away from the kaaba wall also tended to be associated with faster walking speeds .
it is known from other fundamental diagrams , that pedestrians tend to walk faster along a free walkway . as might be expected the walking speeds associated with various factors .
the motion of a single individual at any given time and the direction and speed result in a long list of possible ( and very likely conflicting ) forces and circumstances .
the data taken show that each of the locations and surrounding factors have a significant effect on the behaviour and walking speed of the pilgrims on the mataf area , not forgetting that the age of the pedestrians play a significant role on the tawaf movement and density peaks and jams are caused by pilgrims of age 70 and more . for approximately one half of the location
, the factors examined there also showed an important correlation between pedestrian age , the location and the mean walking speed of the pilgrims .
this funding is consistent with results published by knoblauch @xcite .
the walking speed of pilgrims shows statistically significant variations across a variety of sites , times and environmental conditions ( pedestrian density on the mataf area ) . on the roof of the mosque
the pilgrim density is low and every pedestrian can walk with his desired velocity .
however , the mean walking speed data is explicit by clustered for both pedestrians sex , men and women , independent of the age of the pilgrims are considered . there
exist numerous methods that track the movement of single individuals by inspecting their orientation and limb positions .
this section highlights a real - time system for pedestrian tracking from sequences of high resolution images acquired by a stationary ( high definition ) camera .
the objective was to estimate pedestrian velocities as a function of the local density . with this system
the spatio - temporal coordinates of each pedestrian during the tawaf ritual were established .
processing was done through the following steps : * existing footage was loaded onto a 3d program as a backplate . * from several provided 2d- architectural drawings , a 3d model of the mosque was built . *
a virtual camera was matched in position , rotation and focal length to the original camera so that the features of the 3d - model matched the features positioned on the filmed mosque .
* individual features were identified by eye , contrast is the criterion * we do know that the pilgrims walk on a plane , and after matching the camera we also obtained the height of the plane in 3d - space from our 3d model . *
a point object was placed at the position of a selected pedestrian . during the animation we set multiple animation - keys ( approx every 25 to 50 frames ( equals 1 to 2 seconds ) ) for the position , so that the position of the point and the pedestrian overlay nearly all the time .
* by evolving the point with time we obtained the distance travelled , by measuring the distance from frame to frame .
we also knew the time elapsed from the speed per frame , and hence the speed could be calculated . from figures [ fig : walkingspeed-1 ] and [ fig : walkingspeed-3 ]
we see that the edge of the mataf moves faster than the center , this phenomenon being known as the edge effect . the edge effect occurs when the edges of a crowd move faster than the center of the crowd . the density becomes higher and higher as one moves from the edge of the mataf towards the center .
this phenomenon is explained by the fact that all pilgrims want to be near the kaaba wall . as a result
, we find the density near the kaaba to be the maximum density .
this data can be used in validating of simulation tools . the mean walking speed for a group of pedestrians moving in the pilgrim stream around the kaaba was 1.0816 m / s at the edge of the mataf and it was 0.3267 m / s for the same pedestrians groups moving inside the mataf .
these findings agree well with the statistical results discussed in a previous section .
one of the must - have results is to compare the mean values and variances of walking speeds in both observations and simulation results .
a distinction will be made for walking speeds inside and outside of the mataf platforms .
we made a comparison between our plots derived from the video observation and the fundamental diagrams of ( cf .
[ fig : comparison-1 ] ) : * walking speeds : * * on the edge of the mataf ( free flow speed ) where the pedestrian density is lower than 3 persons / m@xmath9 . * * on the center of the mataf . * * on the mataf inside near the kaaba wall where the pedestrian density attains extreme levels ( 8 - 9 persons / m@xmath9 ) . ) .
average of the local speeds @xmath15 as a function of the local density @xmath10 .
our own data are shown as red points .
the blue points correspond to the data obtained by ( pm ) @xcite .
the difference in velocity at lower densities can be explained by the fitness level of pedestrians . ]
all well - known fundamental diagrams predict the same behaviour and have the same properties : speed decreases with increasing density .
so the discussion above indicates there are many possible reasons and causes for the speed reduction .
for example there is a linear relationship between speed and the inverse of the density for pedestrians moving in a straight way @xcite .
however the pedestrian walking speed can be affected by internal and external factors ( such as the amount of pedestrian inflow and outflow as well as the configuration of the infrastructure ) not to forget the physiology of the human body .
it is found that individuals walk faster in outdoor facilities than in corridors @xcite . according to predtechenskii and milinskii ( pm ) the average walking speed depends on the the walking facility @xcite . in other circumstances
weidmann confirmed a linear relationship between the step size length of walking pedestrians and the inverse of the density @xcite .
the small step size means low pedestrian velocity , caused by reduction of the available space with increasing density .
the discussion above shows that there are many possible factors influencing the fundamental diagram . to identify these factors ,
it is necessary to exclude as many influences of measurement methodology and short range fluctuations from the data .
figure [ fig : comparison-1 ] shows the average local speed @xmath16 as a function of the local density @xmath10 half - hour after mid - day prayer ( t = t@xmath17 ) .
our own data is shown as red points .
the blue points correspond to the milinski fundamental diagram .
moreover investigation data analysing the mataf area represented by blue points in figure [ fig : comparison-1 ] and showed that a reduction of the available navigation space illustrates the causes responsible for the speed reduction with density in pedestrian movement .
the small deviation in pedestrian walking speed at lower density can be explained by the fitness level of the pedestrian .
in the literature , there is a large number of approaches on detection and tracking of moving objects from video images .
spatio - temporal analysis has , in the past , been used to recognize walking persons , where subspaces in the video are treated as spatio - temporal volumes @xcite .
application of a fourier transform to this data can then identify data relating to movement across the volume .
this approach allowed pedestrian trajectories to be reconstructed from video with high precision , taking advantage from the methods and the high developed computational technology . the common approach to detect movement
is to produce comparison images ( an image representing the different details between two images ) since this is computationally efficient @xcite .
these comparison images can then be computed further to estimate movement vectors that describe the motion of drop - shaped objects captured in the respective images .
murakami and wada demonstrate another method , filing the difference frame , and instead compare the properties of drops identified in consecutive frames @xcite .
a drop that is close to the position of a drop in a previous frame , and shares similar dimensions , is likely to refer to the same figure .
motion vectors are also used to find drop segmentation , which are subsequently merged or separated for the purpose of analysis .
the same approach is applied to a 2d image to determine movement in 3d space .
extrapolating the movement of pedestrians in 3d space from a 2d image allows for a far greater understanding of the interactions between entities , but does require exceptional calibrations of equipment for complete accuracy .
the murakami and wada approach can be used to analyse low - quality video streams due to the frame - differencing algorithm and some trigonometry .
determining 3d motion does require precise knowledge of the angle and position of the camera , in addition to the basic topology of the scene being analysed .
but 2d paths are easy to identify without these details , ( see fig . [
fig : pilgrimspaths-2 ] ) . in figure
[ fig : pilgrimspaths-2 ] we show the path of individuals within the crowd .
one clearly recognizes that the movement around the kaaba is not a circle movement .
the tracking of a single individual in the pilgrim stream indicates some oscillation movement around the main path of the individual .
it is caused by the physical repulsive and attractive forces acting on the individual .
physical forces become important when an individual comes into physical contact with another individual / obstacle .
when a local density of 6 persons per square meter is exceeded , free movement is impeded and local flow decreases , causing the outflow to drop significantly below the inflow .
this causes a higher and higher compression in the crowd , until the local densities become critical in specific places on the mataf platform .
in the mataf everything is dense and we have a compact state . the pilgrims have body contact in all directions and no influence on their movement ; they float in the stream .
this forms structures and turbulences in the flow .
these turbulences can be well observed in our video recording .
density and velocity can also be seen .
these observed hajj rituals , especially the mataf , showed some critical points in the motion of the pilgrims that we had not paid much attention to before .
for example : the edge effect , density effect , shock - wave effect etc . , and phenomena like these influence the restraint of the motion and are very important to be considered .
our video analysis shows that the pedestrian density decreases with the distance from the kaaba wall , cf .
figures [ fig : densitydistribution-5 ] , [ fig : density-1 ] , and [ fig : densitymap-1 ] .
it is the same as the real behaviour of pilgrims on the mataf ritual ( all pilgrims want to be near to the kaaba wall ) .
our video analysis about the mataf area indicates that , even at extreme densities , the average local speeds and flows stay limited .
this extremely high local density causes forward and backward moving shock - waves , which could be clearly observed in our video .
we can see a kind of oscillation on the pilgrims paths around the kaaba , this oscillation is caused by shock - waves and is affected by the repulsive forces between the pedestrians in high density crowds ( see fig .
[ fig : pilgrimspaths-2 ] ) .
one of the significant challenges in the planning , design and management of public facilities subject to high density crowd dynamics and pedestrian traffic are the shortcoming in the empirical data .
the collected data concerning crowd behaviour using different techniques ( image processing ) and analysis of ordered image sequences obtained from video recording is increasingly desirable in the design of facilities and long - term site management .
we have investigated the efficiency of a number of techniques developed for crowd density estimation , movement estimation , critical places and events detection using image processing . in the above sections and within this investigation
we have presented techniques for background generation and calibration to improve the previously developed simulation model . even though extracting information about human characteristics from video recording may still be in its infancy , it is important to mention that the field of human motion analysis is large and has a history traced back to the work of hoffman and flinchbaugh @xcite . in the field of pedestrian detection techniques , moreover in the big area of computer vision , many problems have accumulated . in the human motion analysis , and
also in the problem of the detection of moving objects , remain other problems , namely to recognize , categorize , or analyse the long - term pattern of motion .
the inspection of the literature in the last decade indicates increasing interest in event detection , video tracking , object recognition , because of the clear application of these technologies to problems in surveillance .
recently many methods have been developed to extract information about moving object like speed and density .
almost all these systems require complex intermediate processes , such as reference points on the tracked objects or the image segmentation .
one limitation of this current system is that the detection failures for these intermediates will lead to failure for the entire system .
improvement of an algorithm to be able to reproduce traffic flow and to help in the microscopic pedestrian data collection is very essential .
moreover the automatic video data collection will highly enhance the achievement of a system for higher pedestrian traffic densities .
i would like to express my sincerest thanks and gratitude to prof .
g. wunner for a critical reading of the manuscript , for his important comments and suggestions to improve the manuscript .
many thanks to dr .
h. cartarius for his support during writing this work . | in this paper we present a number of methods ( manual , semi - automatic and automatic ) for tracking individual targets in high density crowd scenes where thousand of people are gathered .
the necessary data about the motion of individuals and a lot of other physical information can be extracted from consecutive image sequences in different ways , including optical flow and block motion estimation .
one of the famous methods for tracking moving objects is the block matching method .
this way to estimate subject motion requires the specification of a comparison window which determines the scale of the estimate . in this work we present a real - time method for pedestrian recognition and tracking in sequences of high resolution images obtained by a stationary ( high definition ) camera located in different places on the haram mosque in mecca .
the objective is to estimate pedestrian velocities as a function of the local density.the resulting data of tracking moving pedestrians based on video sequences are presented in the following section . through the evaluated system
the spatio - temporal coordinates of each pedestrian during the tawaf ritual are established .
the pilgrim velocities as function of the local densities in the mataf area ( haram mosque mecca ) are illustrated and very precisely documented .
tracking in such places where pedestrian density reaches 7 to 8 persons / m@xmath0 is extremely challenging due to the small number of pixels on the target , appearance ambiguity resulting from the dense packing , and severe inter - object occlusions .
the tracking method which is outlined in this paper overcomes these challenges by using a virtual camera which is matched in position , rotation and focal length to the original camera in such a way that the features of the 3d - model match the feature position of the filmed mosque . in this model
an individual feature has to be identified by eye , where contrast is a criterion .
we do know that the pilgrims walk on a plane , and after matching the camera we also have the height of the plane in 3d - space from our 3d - model .
a point object is placed at the position of a selected pedestrian . during the animation we set multiple animation - keys ( approximately every 25 to 50 frames which equals 1 to 2 seconds ) for the position , such that the position of the point and the pedestrian overlay nearly at every time . by combining all these variables with the available appearance information
, we are able to track individual targets in high density crowds .
keywords : pedestrian dynamics , crowd management , crowd control , objects tracking . + 2 |
* description of the experiment * a voltage applied to the schottky top gate was used to tune the qd into stable charge configurations with the ground state being singly positively charged , neutral or singly negatively charged @xcite .
single electrons were injected into the qd from the fermi reservoir , whereas in absence of a hole reservoir in the device , single hole injection was ensured by the presence of a weak non - resonant laser . the sample was cooled in a bath cryostat to liquid helium temperature ( 4.2 k ) and finite magnetic fields were applied in faraday or voigt configuration , parallel or perpendicular to the sample growth axis @xmath11 , respectively .
resonant absorption of the neutral exciton @xmath2 as well as positive and negative trions , @xmath4 and @xmath3 , was probed with a tunable narrow - band laser in differential transmission spectroscopy @xcite .
absorption spectra were recorded by setting the gate voltage or the laser energy to a specific detuning @xmath78 , waiting for a time @xmath57 and monitoring the transmission signal with a lock - in amplifier .
after each measurement , the voltage / laser energy was changed by a discrete detuning step @xmath56 . in finite magnetic fields , qds in both samples
a and b showed pronounced dragging .
qds in sample a , for example , exhibited dragging of both @xmath2 and @xmath3 in the range of tens of @xmath1ev ( fig .
[ fig - x ] ) in contrast to qds of sample b with sub 10 @xmath1ev scale for both trions ( fig .
[ fig - xnxp ] ) .
the thickness of the tunnel barrier strongly affects electron spin exchange with the fermi reservoir via co - tunneling @xcite and thus the efficiency of dnsp @xcite .
in particular , in sample b the dragging widths of @xmath3 are reduced due to strong spin pumping at magnetic fields exceeding @xmath12 @xcite . in voigt geometry
the spin pumping leads to a significant reduction in the transmission contrast and consequently in the dragging efficiency even at the edge of the charging plateau .
however , the @xmath2 of the same qd from sample b exhibited a @xmath79 dragging width , consistent with the findings from sample a. for @xmath4 the reduced dnsp range is consistent with the presence of a non - resonant laser which not only injects holes into the ground state of @xmath4 but also non - geminate electron spins opening up an additional nuclear spin decay channel . the nuclear spin polarization is therefore reduced for ( photo - generated ) single - hole - charged initial states .
the effective non - collinear hyperfine interaction hamiltonian stems from the fact that the quadrupolar interaction hamiltonian for a nuclear spin with strain axis tilted by an angle @xmath80 from the @xmath11-axis ( in the @xmath81 plane ) @xmath82 \ ; \label{eqn : quadrupolar}\end{aligned}\ ] ] does not commute with the dominant @xmath83 term of the fermi - contact hyperfine interaction @xmath84 . to obtain an analytic expression for @xmath27
, we assume @xmath85 and use a schrieffer - wolff ( sw ) transformation to obtain @xmath86 where @xmath87 \label{eqn : quadrupolar - hyperfine } \;,\end{aligned}\ ] ] with @xmath15 . in @xmath88
we have only kept the terms that describe processes which leave the electron spin - state unchanged , since contributions that flip the electron spin will be negligible at high external fields .
finally , we note that even for large @xmath13 , the dominant role of flip - flop terms of fermi - contact hyperfine interaction is to induce indirect interaction between the qd nuclei @xcite : the primary effect of this interaction , in the presence of fast optical dephasing of the electonic spin resonance , is to ensure that the nuclear spin population assumes a thermal distribution on timescales fast compared to the polarization timescale determined by @xmath27 . in this limit
, the dynamics due to @xmath88 will be indistinguishable from that described by @xmath33 .
averaging over the distribution for nuclei that lie within the gaussian qd electron wavefunction , we obtain for cations ( in and ga with 9/2 and 3/2 nuclear spins , respectively , and @xmath89ev , @xmath90ev @xcite ) @xmath91 and for anions ( as with 3/2 nuclear spin and @xmath92ev @xcite ) @xmath93 . for a fully polarized in@xmath30ga@xmath31as
system we determine the value for the maximum overhauser splitting due to the non - collinear hyperfine coupling as @xmath94ev and obtain an average value of @xmath95ev with @xmath96 .
next , we discuss the derivation of the rate equation for the neutral exciton blue zeeman transition .
we consider the limit of a large external magnetic field where @xmath97 , with @xmath98 ( @xmath99 , @xmath19 are the electron and nuclear zeeman energies , @xmath44 the laser rabi frequency and @xmath45 the radiative decay rate ) . in this limit ,
all nuclear spin - flip processes , including those described by @xmath33 are energetically forbidden to first order in perturbation theory .
eliminating @xmath33 by a sw transformation we arrive at the following correction terms to the laser - exciton coupling @xmath100 with @xmath101 . here ,
@xmath102 and @xmath103 denote the exciton and vacuum state , respectively .
application of the same sw transformation to the liouvillian term leads to nuclear - spin - flip assisted spontaneous emission terms with maximum rate @xmath104 . in the limit @xmath105 of interest , the denominator in eq .
( [ eqn : nc - laser ] ) should be modified to take into account broadening of the excitonic spin states due to spontaneous emission .
the rates for spin - flip raman scattering processes arising from the fermi - contact hyperfine interaction take place at a rate @xmath106 ; given that @xmath107 and @xmath108 , we conclude that the latter processes will take place at a rate that is @xmath109 times slower .
s and @xmath110ev ) with and without spin - flip raman scattering processes according to eqn.s [ eqn : iz - roh - som ] and [ eqn : iz - som ] , respectively . the slight asymmetry to positive laser detunings is a result of directional dynamic nuclear spin polarization stemming from spin - flip raman processes .
( b ) comparison between experimental spectra ( dark cyan ) and results of the simulation ( blue ) for sequential data acquisition ( @xmath111 s and @xmath112ev ) : each data point of the spectra was obtained by ( i ) erasing the nuclear spin polarization in a voltage region with strong co - tunneling , ( ii ) subsequently establishing a finite laser detuning , and ( iii ) integrating for the time @xmath113 at this specific detuning .
the lorentzian spectra in ( a ) and ( b ) are calculated with @xmath114ev and shown in black for reference . ]
t for the neutral exciton @xmath2 ( blue zeeman branch : closed circles ; red zeeman branch : open circles ) and the negative trion @xmath3 ( open triangles ) in sample a. the dragging width is normalized to the power - broadened total linewidth @xmath115 , the laser power is expressed as @xmath116 ( with experimentally determined radiative decay rates @xmath117 ns@xmath118 for @xmath2 and @xmath119 ns@xmath118 for @xmath3 ) .
the results of the model are shown by the solid line for @xmath116 and dashed line for @xmath120 .
the inset shows on the same abscissa scale in red and blue the corresponding degree of nuclear spin polarization accumulated in a forward scan at initially negative and positive laser detunings , respectively .
small but finite degree of nuclear spin polarization at high laser powers stems from directional spin - flip raman scattering processes .
( b ) maximum dragging width of @xmath2 ( circles ) as a function of magnetic field for @xmath121 .
the sub - linear monotonic increase of the maximally achieved dragging width is reproduced by the simulations ( solid line ) .
the linearly increasing electron zeeman energy @xmath122 is also shown ( dashed line ) . ] to obtain a quantitative prediction , we consider the rate equation @xmath123 where @xmath124 is the rate at which nuclear - spin - flip assisted spontaneous emission , leading to pure nuclear spin diffusion , takes place . here , as well as in the following equation , @xmath125 .
the rate equation yields symmetric dragging ( dashed spectra in fig .
[ fig - som - spectra](a ) ) and anti - dragging to either side of the resonance for the blue and red zeeman branches , respectively , qualitatively similar to the results of yang and sham @xcite . taking into account uni - directional spin - flip raman scattering processes that arise from the fermi - contact hyperfine interaction , we arrive at a refined rate equation model : @xmath126 with @xmath127 spin - flip raman scattering processes at rate @xmath128
give rise to the asymmetry in the spectra for forward and reverse scan directions ( compare solid and dashed spectra in fig . [ fig - som - spectra](a ) ) , in agreement with experimental findings .
the argument for the asymmetry holds when the nuclear spin polarization is erased before a sudden jump to a finite detuning and subsequent build - up of dnsp @xcite : locking of the resonance extends further for positive laser detunings ( fig .
[ fig - som - spectra](b ) ) .
moreover , spin - flip raman scattering processes ensure small but finite nuclear spin polarization at high laser powers , as shown in the inset of fig .
[ fig - som - power](a ) .
the normalized absorption spectra depicted in fig .
[ fig - som - spectra](a ) are calculated from steady state solutions of eqn.s [ eqn : iz - som ] and [ eqn : iz - roh - som ] with the following parameters : @xmath68ev from the radiative decay rate @xmath129 ns determined from saturation @xcite , @xmath69 , @xmath70 t , step size @xmath71ev and dwell time @xmath72 s , as used in the experiments , and @xmath73 , @xmath74 , @xmath75 , @xmath130ev .
intrinsic decay of the nuclear spin polarization is negligible for the ground state of @xmath2 @xcite . here
, we omitted the unbalanced telegraph noise in the resonance condition used to calculate the spectra in fig . 1 of the manuscript .
this jitter in the resonance condition with an amplitude of 0.5 @xmath1ev ( smaller than the linewidth ) and timescales longer than @xmath57 was included in simulations in order to account for the asymmetry observed experimentally for the two scan directions .
it is consistent with the spectral fluctuations in resonant qd spectroscopy @xcite .
based on experimental observations for the qds in sample a , it is reasonable to assume these fluctuations to be unbalanced with a small weight on the higher energy side of the resonance .
however , comparing the observed traces for different qds in different samples reveals that the spectral jitter can actually appear on either side of the resonance .
the dependence of the dragging width on the magnitude of the laser power and the external magnetic field provides further confirmation of the model .
[ fig - som - power ] shows how the dragging width evolves as a function of laser power and magnetic field .
the effect of dragging is inhibited at low incident powers , increases until reaching a maximum below the saturation at @xmath131 , and vanishes in the limit of high excitation powers ( fig .
[ fig - som - power](a ) ) .
this is consistent with the prediction of eq.s [ eqn : iz - som ] and [ eqn : iz - roh - som ] : the maximum dragging width is expected at @xmath132 for non - vanishing @xmath133 and @xmath134 .
both @xmath2 and @xmath3 reveal the same dependence when the dragging width on the ordinate is normalized to the total linewidth @xmath135 and the abscissa is expressed is units of @xmath44/@xmath45 .
the results of the model reproduce our experimental findings ( solid line in fig .
[ fig - som - power](a ) ) predicting @xmath136 of nuclear spin polarization at maximum ( inset of fig .
[ fig - som - power](a ) ) .
the model overestimates the dragging width at powers below saturation but gives perfect agreement above saturation for @xmath137 .
we speculate that the scaling factor of 2 stems from the line broadening @xmath138 that we typically find in our samples as a result of spectral fluctuations @xcite .
the monotonic sublinear increase of the dragging width with magnetic field , as measured on @xmath2 close to saturation , is clearly reproduced by our model ( solid line in fig .
[ fig - som - power](b ) ) and provides further confirmation for the quantitative nature of our analysis . | resonant optical excitation of lowest - energy excitonic transitions in self - assembled quantum dots lead to nuclear spin polarization that is qualitatively different from the well known optical orientation phenomena . by carrying out a comprehensive set of experiments
, we demonstrate that nuclear spin polarization manifests itself in quantum dots subjected to finite external magnetic field as locking of the higher energy zeeman transition to the driving laser field , as well as the avoidance of the resonance condition for the lower energy zeeman branch .
we interpret our findings on the basis of dynamic nuclear spin polarization originating from non - collinear hyperfine interaction and find an excellent agreement between the experimental results and the theoretical model .
the basic principles of optical nuclear spin orientation in solids have been studied extensively in bulk semiconductors @xcite and attracted a revived attention by recent optical studies of semiconductor quantum dots ( qds ) .
dynamic nuclear spin polarization ( dnsp ) in self - assembled ingaas qds has been reported for quasi - resonant @xcite and non - resonant excitation @xcite . on the basis of these experiments and
related theoretical studies , a comprehensive picture of unidirectional optical orientation of qd nuclear spins effected by light - polarization selective pumping was developed .
early experiments carried out on positive and negative trions @xcite as well as neutral excitons @xcite had been used to demonstrate bistability of dnsp as a function of magnetic field or incident laser power and polarization @xcite , and to study nuclear spin buildup and decay dynamics @xcite . in stark contrast to non - resonant excitation however
, bidirectional nuclear spin orientation independent of photon polarization was observed in resonant laser scattering of elementary transitions in neutral @xcite and negatively charged qds @xcite .
a particularly striking feature of resonant dnsp using the higher energy zeeman transition at external magnetic fields exceeding @xmath0 t is the flat - top absorption spectra , stemming from active locking of the qd resonance to the laser frequency @xcite .
remarkably , neutral and negatively charged qds showed similar spectral signatures in resonant spectroscopy despite substantially different energy level diagrams : for both charge states , the locking of the coupled electron - nuclear spin system to the incident laser ( dragging ) was observed over tens of @xmath1ev detunings to either side of the resonance @xcite . in this letter
, we carry out a comprehensive experimental and theoretical analysis of dragging in resonantly driven qd transitions .
we develop a microscopic model based on the effective non - collinear hyperfine coupling that was first proposed by ref .
@xcite to explain nuclear spin relaxation in self - assembled qds .
our experiments demonstrate that the nature of resonant dnsp depends drastically on whether the blue ( higher energy ) or the red ( lower energy ) zeeman transition is resonantly excited ; while the blue transition exhibits locking of the qd resonance to the incident laser , nuclear spin polarization ensures that the resonance condition is avoided for the red transition @xcite .
we also find that while the frequency range over which blue zeeman transition locking takes place varies from qd to qd , the dependence of the corresponding dragging width on laser power , scan speed and the magnetic field is similar for all qds .
a key requirement for dragging is the presence of an unpaired electron spin with a long spin - flip - time , either in the initial or the final state of the optical transition ; this condition is satisfied by fundamental neutral ( @xmath2 ) , single - electron ( @xmath3 ) and single - hole ( @xmath4 ) charged qd transitions .
the overhauser field @xcite experienced by this unpaired electron facilitates the feedback that modifies the qd transition energy . however , whether or not this feedback leads to _ resonance seeking _ ( as in the blue zeeman branch ) or _ resonance avoiding _ ( as in the red zeeman branch ) excitations depends on the spin orientation of the electron that couples to the incident laser field .
we studied individual ingaas qds embedded in a field effect device @xcite .
two samples distinct by the thickness of the tunnel barrier between the heavily @xmath5-doped back contact and the qd layer ( 25 nm and 35 nm in samples a and b , respectively ) were employed to probe the fundamental exciton transitions in resonant laser scattering experiments at 4.2 k @xcite .
representative spectra measured on the neutral exciton @xmath2 in sample a subjected to a moderate magnetic field of @xmath6 t are shown in the upper panels of fig .
[ fig - x](c ) and ( d ) .
the spectra instantly reveal drastic departures from a two - level lorentzian with resonance frequency @xmath7 : the blue zeeman optical transition of @xmath2 shows flat - top absorption ( fig . [ fig - x](c ) ) , also reported earlier for the negative trion in faraday @xcite and voigt @xcite configurations .
in contrast , the spectral shape of the red zeeman resonance is triangular ( fig .
[ fig - x](d ) ) with maximum contrast that is a factor of @xmath8 lower than its blue counterpart . from direct comparison
it becomes apparent that the blue transition is locked to the laser at frequency @xmath9 and can be dragged to positive and negative laser detunings @xmath10 by tens of @xmath1ev , dependent on the scan direction ( grey and blue spectra in fig .
[ fig - x](c ) ) .
in contrast , the red transition avoids the resonance with the laser using dnsp ( fig .
[ fig - x](d ) ) , resulting in a triangular line shape .
t as a function of quantum dot charge and magnetic field orientation : positive and negative trions , @xmath4 and @xmath3 , exhibit characteristic flat - top resonances on the blue zeeman transition and triangular lineshapes on the red zeeman transition ( left and right panels in a and b , respectively ) in magnetic field oriented ( a ) parallel ( faraday ) and ( b ) perpendicular ( voigt ) to the sample growth axis @xmath11 . ]
we systematically measure the same qualitative response for the blue ( red ) zeeman transitions of @xmath2 as well as of both trions , @xmath4 and @xmath3 , in faraday and voigt magnetic field geometries , as demonstrated in fig .
[ fig - xnxp ] .
there are , however , quantitative variations in the efficiency of dnsp from sample to sample and even from dot to dot within one sample @xcite .
the thickness of the tunnel barriers in samples a and b plays a crucial role for electron spin exchange with the fermi reservoir via co - tunneling @xcite and thus for electron - spin pumping at magnetic fields exceeding @xmath12 @xcite as well as the efficiency of dnsp @xcite .
our findings demonstrate that effects of bidirectional dnsp are omnipresent in resonant laser spectroscopy of qd excitons and call for a unified explanation that goes beyond directional dnsp mediated by flip - flop terms of the fermi - contact hyperfine interaction .
obviously , the model should be insensitive to the details of the initial and final qd states , such as charge configuration or the presence of dark exciton states @xcite , yet capture marked signatures and differences in the response of the blue and red zeeman transitions to a near - resonant laser .
recently , yang and sham @xcite proposed that non - collinear hyperfine interaction between heavy - holes and the nuclei , induced by heavy - light - hole coupling , provides an excellent qualitative description of the signatures related to dnsp in resonant laser scattering @xcite . on the other hand ,
recent experiments @xcite demonstrate that non - collinear hyperfine interaction between the electron and the nuclei plays a significant role in determining qd nuclear spin dynamics even in the absence of optically generated holes ; this interaction is induced by large quadrupolar fields in strained self - assembled qds which ensure that nuclear spin projection along @xmath13 is not a good quantum number .
the resulting effective non - collinear interaction between the qd electron and the nuclei is @xcite : @xmath14 with @xmath15 , and @xmath16 , @xmath17 , spin operators of the electron spin and the @xmath18-th nucleus , respectively . here
, @xmath19 denotes the nuclear zeeman energy , @xmath20 the strength of the quadrupolar interaction and @xmath21 is the angle between the major quadrupolar axis of the @xmath18-th nucleus and the @xmath11-axis . for the coupling strength of the electron to the @xmath18-th nucleus we assumed @xmath22 , where @xmath23 is the maximal overhauser field splitting and @xmath24 the number of nuclei . to determine @xmath20 and @xmath21 , we first employed molecular statics with tersoff type force fields @xcite to obtain the realistic structure for more than one million atoms hosting @xmath25 qd nuclei
. the atomistic strain and nuclear quadrupolar distributions are extracted over this relaxed structure @xcite .
fig .
[ fig - strain](a ) shows the distribution of the biaxial strain @xmath26 which is primarily responsible for the nuclear quadrupolar shifts .
based on this distribution , we determine @xmath27 for a line - cut along the qd taken through the center and the @xmath28 $ ] axis , cf . fig .
[ fig - strain](b ) .
averaging over this distribution for nuclei that lie within the gaussian qd electron wavefunction , we obtain @xmath29ev , consistent with @xcite . , which primarily controls the quadrupolar splitting , over the ( 100 ) plane bisecting a truncated coned - shaped quantum dot .
the false color plot illustrates the tensile strain distribution with rapid variations within the quantum dot region arising from the random composition of the in@xmath30ga@xmath31as alloy used in the model .
( b ) the value of @xmath32 evaluated over the line - cut through the monolayer indicated by the solid line in ( a ) .
averaging over this distribution with a gaussian envelope for the electron wavefunction we obtain a value of 0.0062 for cations ( in or ga ) and 0.0424 for anions
( as ) .
] the fact that @xmath33 could explain dragging is at first sight surprising since its dominant effect appears to be nuclear spin diffusion .
however , a careful inspection shows that the same hamiltonian also leads to a small polarization term whose direction is determined by the sign of the optical detuning @xcite . to explain this , we consider the energy level diagrams of @xmath2 in fig .
[ fig - x](a ) and ( b ) , each showing a ladder of two - level quantum systems coupled by nuclear spin - flip processes . here
we adopt a mean - field description of the nuclear spins by neglecting the quantum fluctuations in the overhauser field ( @xmath34 ) and limit ourselves to effective spin @xmath35 nuclei for simplicity . for a given nuclear spin polarization @xmath36 , e.g. @xmath37
, we can label the two level system by the states @xmath38 and @xmath39 .
the ground states @xmath38 differ by an energy @xmath40 .
the corresponding energy differences for the excited states are @xmath41 and @xmath42 for the blue and red zeeman transition , respectively . the transition rate associated with hyperfine - assisted laser coupling
is given for the blue zeeman branch ( fig .
[ fig - x](a ) ) by @xmath43 where @xmath44 is the laser rabi frequency and @xmath45 the radiative decay rate @xcite .
a remarkable feature of @xmath46 is its dependence on the sign of the laser detuning entering through the effective optical detuning @xmath47 : when the incident laser field is red ( blue ) detuned , the transition rate @xmath48 ( @xmath49 ) dominates over @xmath49 ( @xmath48 ) and ensures that the overhauser field increases ( decreases ) .
this directional dnsp will in turn result in a decrease of the effective detuning @xmath50 from @xmath51 to @xmath52 for a red detuned laser and to @xmath53 for a blue detuned laser .
if initially @xmath54 , then dnsp will continue until @xmath55 . in sample
a at 4.5 t as a function of discrete laser detuning step @xmath56 and integration time constant @xmath57 . experimental spectra ( a ) and results of the simulations ( b ) obtained with steps in laser detuning of @xmath58ev at three time constants @xmath57 of 0.2 , 0.5 and 5.0 s ( shown as grey , dark grey and black solid lines , respectively ) .
the parametric plot in ( c ) depicts the dragging width obtained for incremental laser detuning steps @xmath59ev from @xmath60ev ( top , dark blue ) to @xmath61ev ( bottom , red ) at different time constants ( closed circles : experiments , solid lines : simulations ; the parametric increment @xmath5 is given in numbers , the set of data was normalized to the value of the dragging width for @xmath62 at 5 s ) .
the full dynamic range of the experiment is correctly reproduced by the model . ]
while a laser scan across the blue transition leads to a positive feedback of the nuclear spins to ensure locking condition , a scan across the red zeeman line causes an anti - dragging effect . to understand this
, we note that the effective optical detuning in this case is @xmath63 .
the simple sign change in the effective optical detuning renders the exact resonance between the laser field and the red exciton transition an unstable point .
the dnsp that ensues in the presence of a small but non - zero @xmath50 will result in nuclear spin flip processes that increase @xmath64 resulting in pushing the red zeeman transition away from the laser resonance .
the experiments validate these conclusions ( fig .
[ fig - x](d ) and right panel of fig .
[ fig - xnxp ] ) . to obtain a quantitative prediction
, we consider the rate equation @xmath65 which includes nuclear - spin - flip assisted spontaneous emission processes leading to pure nuclear spin diffusion at rate @xmath66 @xcite .
the steady - state solution exhibits bistability due to the nonlinear @xmath36 dependence of the rates @xmath46 . for a given initial laser detuning @xmath67 the solution of the rate equation
yields the steady state nuclear spin polarization @xmath36 and thus the effective optical detuning @xmath50 as established within the integration time @xmath57 of the experiment .
the absorption spectrum is calculated by varying the laser detuning in discrete steps @xmath56 .
the rate equation ( [ eqn : iz ] ) is symmetric with respect to the laser detuning . in order to account for the asymmetry
observed experimentally for the two scan directions , we refined the model by including terms for spin - flip raman scattering processes that arise from the fermi - contact hyperfine interaction as well as an unbalanced telegraph noise in the resonance condition @xcite .
we find excellent agreement between theory and experiment , as demonstrated in fig .
[ fig - x ] .
remarkably , the model also reproduces the dependence of dnsp in resonant laser scattering on key experimental parameters .
fig .
[ fig - speed ] shows how the dragging width evolves as a function of scan speed .
both the detuning step @xmath56 and the waiting time constant @xmath57 used for signal integration after each step contribute to the effective scan speed of the laser detuning : for a given @xmath56 the maximum width increases non - linearly with @xmath57 , as exemplified in fig .
[ fig - speed](a ) for @xmath58ev .
the non - linearity in the functional dependence makes our simulations highly sensitive to the set of parameters that determine the dnsp dynamics ; in particular , it allows us to extract a value for @xmath27 for a given set of @xmath24 and @xmath23 .
fig .
[ fig - speed](c ) demonstrates that the full dynamic range of the experiment is correctly captured with the following set of parameters : @xmath68ev , @xmath69 , @xmath70 t , step size @xmath71ev and dwell time @xmath72 s , as used in the experiments , and @xmath73 , @xmath74 , @xmath75 , @xmath76ev .
the value for the non - collinear hyperfine coupling constant found from simulations is in good agreement with that obtained independently from atomistic calculations and nuclear spin decay measurements @xcite .
the same set of parameters was also used to reproduce the external magnetic field and the laser power dependence @xcite and to calculate the absorption spectra in fig .
[ fig - x ] .
our results establish quadrupolar interaction induced non - collinear hyperfine coupling as the mechanism responsible for resonant bidirectional dnsp that is ubiquitous for self - assembled qds .
an obvious extension of our work will be to carry out similar experiments in interface or droplet qds where quadrupolar interactions are vanishingly small .
we acknowledge financial support from nccr - nanoscience , erc , and the dfg ( sfb 631 ) .
we thank j. m. sanchez , a. badolato , d. schuh and w. wegscheider for growing the two samples used in this work .
we also acknowledge useful discussions with m. atature , n. vamivakas , y. zhang , m. issler and p. maletinsky .
f. meier and b. p. zakharchenya , _ optical orientation _
( north - holland , amsterdam , 1984 ) . c. w. lai , p. maletinsky , a. badolato , and a. imamoglu , phys .
rev .
lett . * 96 * , 167403 ( 2006 ) .
b. eble _ et al .
_ , phys .
rev .
b * 74 * , 081306 ( 2006 ) .
a. tartakovskii _
et al . _ ,
phys .
rev .
lett .
* 98 * , 026806 ( 2007 ) .
t. belhadj _
et al . _ ,
phys .
. lett . * 103 * , 086601 ( 2009 ) .
p. maletinsky , c. w. lai , a. badolato , and a. imamoglu , phys .
rev .
b * 75 * , 035409 ( 2007 ) .
p .- f .
braun _
et al .
_ , phys .
rev .
b * 74 * , 245306 ( 2006 ) .
p. maletinsky , a. badolato , and a. imamoglu , phys .
rev .
lett . *
99 * , 056804 ( 2007 ) .
p. maletinsky , m. kroner , and a. imamoglu , nature phys .
* 5 * , 407 ( 2009 ) . c. latta _ et al . _ , nature phys . * 5 * , 758 ( 2009 ) .
x. xu _
et al . _ ,
nature * 459 * , 1105 ( 2009 ) .
t. d. ladd _
et al . _ ,
phys .
rev .
lett .
* 105 * , 107401 ( 2010 ) .
c .- w . huang and x. hu , phys .
rev .
b * 81 * , 205304 ( 2010 ) . w. yang and l. j. sham , arxiv:1012.0060 ( 2010 ) .
h. drexler , d. leonard , w. hansen , j. p. kotthaus , and p. m. petroff , phys .
rev .
lett . * 73 * , 2252 ( 1994 ) .
a. hgele _
et al . _ ,
phys .
rev .
lett .
* 93 * , 217401 ( 2004 ) .
see supplementary online material for details .
j. m. smith _
et al . _ ,
phys .
rev .
lett . * 94 * , 197402 ( 2005 ) .
m. atatre _
et al . _ ,
science * 312 * , 551 - 553 ( 2006 ) .
m. bayer _ et al .
_ , phys .
rev .
lett .
* 82 * , 1748 ( 1999 ) .
experimentally determined values of hole - hyperfine interaction would suggest that the resulting nuclear spin - flip rates would be an order of magnitude too small to explain the relative insignificance of the directional reverse overhauser effect .
in addition , the large apparent variation in heavy - light - hole mixing in qds as indicated by the measured in - plane hole @xmath77-factors would suggest that dragging features would change appreciably from qd to qd , if the mechanism was due to hole - hyperfine interaction . c. latta , a. srivastava , and a. imamoglu , phys .
rev .
lett . * 107 * , 167401 ( 2011 ) .
d. powell , m. a. migliorato , and a. g. cullis , phys .
rev .
b * 75 * , 115202 ( 2007 ) . c. bulutay , in preparation . by carrying out a complete numerical analysis , we have confirmed that the saturation of @xmath46 is correctly taken into account in eq . ( 2 ) . |
_ experiments . _ recent experiments in low - dimensional quantum magnets @xcite have revealed remarkably strong thermal transport anomalies associated with the low - dimensional spin degrees of freedom
. in particular , an anisotropic thermal conductivity , comparable in magnitude to that of metallic systems , was observed in quasi - one - dimensional ( 1d ) spin-@xmath8 chain and ladder compounds with a large part of the heat current attributed to magnetic excitations.@xcite in the quasi-2d and some quasi-1d materials various other effects were observed including strong increase of the thermal conductivity by a modest magnetic field@xcite and enhanced scattering due to suppression of the gap in the spin - gaped materials.@xcite the thermal transport anomalies in layered cuprates associated with the magnetic and/or stripe excitations have also been found.@xcite both the strength and the temperature range for these new effects are very different from their counterparts in 3d magnetic materials , where the heat transport by well - defined magnetic excitations exist only well below the corresponding 3d ordering transition temperature.@xcite note that the thermal transport in low - dimensional magnets has been studied before@xcite and the qualitative reasons for that difference has been understood . in 2d systems
magnetic excitations can be very well defined in a paramagnetic phase because of the large in - plane magnetic correlation length@xcite and in 1d systems the soliton - like excitations can exist without any long - range order.@xcite _ theories .
_ however , only recently a considerable progress in the theoretical understanding of the transport properties of 1d quantum spin systems has been made.@xcite the main focus of these recent studies has been on the relationship of the spin transport and conservation laws,@xcite specifically on the possibility of an ideal conducting state in one - dimensional integrable and non - integrable systems.@xcite while this problem is of significant interest , the importance of the spin - phonon and spin - impurity couplings , which break down the integrability of the underlying spin - only models , has also been discussed.@xcite in particular , it was obtained within the memory matrix formalism that the interplay of umklapp scattering and spin - phonon coupling leads to the exponential temperature dependence of the thermal conductivity @xmath9 , with @xmath10 proportional to the phonon s debye energy @xmath3.@xcite _ boltzmann equation approach . _ in the present work@xcite we consider the problem of anomalous heat transport in quasi-1d spin - chain systems under a somewhat different prospective .
first , we derive microscopically a model of 1d spin - boson excitations interacting with the 3d phonon environment and impurities , where the spin - boson representation of the 1d heisenberg model is obtained by performing the standard jordan - wigner transformation followed by bosonization.@xcite it is assumed that the spin - boson velocity is large in comparison with the phonon velocity , @xmath11 .
the limit @xmath12 ( @xmath13 ) corresponds to the experimental situation in sr@xmath0cuo@xmath1 , srcuo@xmath0 , ( la , ca , sr)@xmath14cu@xmath15o@xmath16 , and other cuprate materials@xcite where @xmath17k and @xmath18 . this large difference in spin and phonon energy scales is also responsible for the fact that , experimentally , the `` spin peak '' in thermal conductivity occurs at the temperatures well above the `` phonon peak''.@xcite with this assumption and in the limit of weak spin - lattice coupling we solve the boltzmann equation for the spin - boson distribution function and find the spin - phonon relaxation time @xmath19 . _
normal vs umklapp .
_ why @xmath19 is important ?
generally , the relaxation of the heat current and the resulting finite thermal conductivity should be due to the umklapp processes or any other processes which do not conserve momentum . since the characteristic `` bandwidth '' for the spin excitations in the real 1d spin chains materials of interest is very large ( @xmath17k ) in comparison with the experimental temperature range , the umklapp scattering of spin excitations on themselves is strongly suppressed ( as @xmath20 ) .
instead , by considering the limit @xmath21 , we propose that the leading relaxation mechanism at not too low temperatures is the two - stage , bottle - neck process of _
( i ) _ transferring momentum from the spin system to phonons , and _
( ii ) _ subsequent dissipation of the phonon momentum via an umklapp process or impurity scattering .
the central idea is that the excess momentum of a spin - boson waits the longest time to get transfered to phonons , but once it is transferred it relaxes quickly . in other words , the phonon relaxation time @xmath22 due to
phonon - phonon scattering is much shorter than the spin - phonon relaxation time @xmath19 , i.e. @xmath23 . as a result , the relaxation rate of such a two - stage process is determined by the `` normal '' spin - phonon scattering rate @xmath24^{-1}\approx\tau_{sp}^{-1}$ ] .
the inequality @xmath25 can be justified qualitatively with the help of the following argumentation .
first , the spin - lattice coupling is weak .
second , since @xmath26 the phonon states have a significant thermal population at the temperatures where the spin contribution to the transport is substantial .
furthermore , there is also a kinematic argument in favor of @xmath23 that results from our analysis .
the leading contribution to the transport comes from the spin bosons with a small momentum @xmath27 .
we find that the most effective scattering mechanism for such bosons is an absorption of a `` thermal '' phonon , that is a phonon whose energy is of the order of @xmath28 and momentum is of the order @xmath29 . for finite @xmath28
such phonons have finite relaxation time @xmath30 .
we will show that the spin bosons in question have a divergent relaxation time @xmath31 .
therefore , for spin bosons with sufficiently small @xmath32 s the relevant phonon relaxation time @xmath30 is shorter than the spin - phonon relaxation time @xmath19 not only because of the weak spin - phonon coupling , but also due to the smallness of the characteristic momentum of the spin - boson , @xmath33 . we have performed a detailed study of the kinematics of the spin - phonon scattering processes and obtained the corresponding transport relaxation rate @xmath34 for @xmath35 , and @xmath36 for @xmath37 , @xmath38 is a constant related to the spin - phonon coupling , @xmath39 .
the characteristic power of @xmath40 in the relaxation rate can be traced back to the fact that spin excitations are confined to lower dimensions than phonons .
the latter implies that it is only the momentum along the chains which is conserved in the spin - phonon scattering .
this , in turn , leads to a larger scattering space for phonons and makes relaxation rate of the spin bosons @xmath41 .
note that for the problem of the phonon - phonon scattering in 3d the phonon scattering rate is @xmath42 because the scattering space is restricted by the momentum conservation in all three dimensions.@xcite the temperature dependence of the relaxation rate is due to the bosonic nature of both the spin excitations and the phonons .
thus , the spin - phonon relaxation rate obtained in this work should apply to other low - dimensional spin systems as well . _ infrared divergence . _ however
, one can find that the considered scattering mechanism becomes too weak at low energy and ineffective for the dissipation of the low - energy spin bosons .
namely , within the formalism of boltzmann equation , @xmath43 leads to the infrared diverging thermal conductivity @xmath44 , @xmath45 , the situation familiar from the phonon thermal conductivity in 3d insulators.@xcite in that latter 3d phonon problem the regularization of such a divergence is non - trivial and involves a consideration of the higher - order phonon processes,@xcite degeneracy of phonon branches at high symmetry points,@xcite or scattering on boundaries .
_ impurities .
_ what physical effect can render @xmath46 finite in our case ?
in contrast with the 3d systems , @xcite because of the 1d nature of spins , even a weak impurity potential will have a dramatic effect on the low - energy spin bosons .
the impurities generate a relevant interaction in the renormalization group ( rg ) sense .
that is , they scatter low - energy excitations very well destroying their quasi - ballistic propagation .
below the so - called kane - fisher scale @xmath47 the scattering is very strong and leads to the localization of 1d excitations.@xcite on the other hand , at the temperatures well above this scale @xmath48 the impurity scattering can be analyzed perturbatively . as a consequence of such an analysis we find that impurities result in the momentum - independent scattering rate @xmath49 , @xmath50 being the impurity concentration .
total scattering rate @xmath51 is finite at @xmath52 , which makes @xmath46 finite .
_ results . _ altogether , the combined effect of impurities and the spin - phonon scattering leads to the following spin - boson thermal conductivity behavior : _ ( i ) _ in the low - temperature , impurity - scattering dominated regime @xmath53 , _ ( ii ) _ in the spin - phonon scattering dominated regime , intermediate temperatures @xmath54 , @xmath55 , and _
( iii ) _ in the spin - phonon scattering dominated regime , higher temperatures @xmath56 , @xmath57 . here
the temperature @xmath58 corresponds to the maximum in @xmath59 .
it is also the crossover temperature between the impurity- and phonon - scattering dominated regimes .
we find that this temperature behavior of the thermal conductivity agrees very well with the available experimental data for the spin - chain material sr@xmath0cuo@xmath1.@xcite the thermal conductivity for the zig - zag chain srcuo@xmath0 and the spin - ladder compounds ( ca , la , sr)@xmath14cu@xmath15o@xmath16,@xcite whose underlying spin models are different from the ones considered in this work , is briefly discussed .
we analyze the impurity concentration dependence of several quantities . as a result
we predict an unusual behavior of : _
( i ) _ the crossover temperature @xmath60 , _ ( ii ) _ the spin - boson thermal conductivity maximum value @xmath61 , and _ ( iii ) _ the asymptotic value of @xmath46 at @xmath62 : @xmath63 .
_ outline . _
our paper is organized as follows . in sec .
[ hamiltonian ] we introduce the spin - chain hamiltonian and derive the spin - phonon and spin - impurity interaction terms . in sec .
[ phonons ] we find the spin - boson relaxation time due to scattering off the phonons . in sec .
[ impurities ] the impurity contribution to the spin - boson relaxation time is calculated . in sec .
[ kappa ] the thermal conductivity _ vs _
temperature is obtained and the results are compared to experimental data . in section [ kappa ]
we also put forward several theoretical predictions for the impurity concentration dependence of different quantities and suggest further experiments . in sec .
[ meanfp ] we discuss the notion of the mean free path and its applicability to the transport by the long - wavelength spin bosons .
we conclude by sec .
[ conclusions ] which contains the discussion of our approach and approximations . technically involved details concerning the spin - phonon collision integral and impurity scattering are described in appendix [ app_a ] and appendix [ app_b ] , respectively .
_ spin - chain hamiltonian .
_ the hamiltonian of a single heisenberg antiferromagnetic spin-@xmath64 chain is : @xmath65 this hamiltonian , after the jordan - wigner transformation , can be expressed in terms of fermionic operators @xmath66 as follows : @xmath67 it is convenient to bosonize this hamiltonian . to do that one introduces chiral fermionic fields @xmath68 such that @xmath69 .
these fields can be rewritten in terms of spin boson field @xmath70 and its dual field @xmath71 as @xmath72 using @xmath73 one writes the spin - chain hamiltonian as : @xmath74 where @xmath75 , @xmath76 is the linear size of the system , the luttinger - liquid parameter @xmath77 is equal to @xmath78 , and the spin - boson velocity @xmath79 is given by @xmath80 where @xmath81 is the chain lattice constant .
furthermore , the spin - boson variables are redefined according to the rules : @xmath82 then we introduce the creation and annihilation operators for the field @xmath83 : @xmath84 which diagonalize the hamiltonian @xmath85 , eq .
( [ h00 ] ) : @xmath86 for details of this procedure see ref . .
we would like to note that the hamiltonian for the spin-@xmath64 chain contains other terms beyond the ll form above .
these terms do not lead to the finite heat conductivity by themselves and thus are not included in our consideration .
the role of such terms in the presence of phonons via a phonon phonon - assisted umklapp process has been considered elsewhere.@xcite _ spin - phonon hamiltonian . _ since the superexchange @xmath2 is a function of inter - site separation , the lattice vibrations are able to modify it .
this mechanism will , therefore , couple phonons to the spin degrees of freedom . we can account for this coupling by expanding @xmath2 in gradients of inter - site distance @xmath87 : @xmath88 where @xmath89 is the displacement along the chain .
this translates into the following interaction hamiltonian : @xmath90 where @xmath91 is a spin - phonon coupling constant .
the atomic displacement vector @xmath92 can be expressed in terms of phonon creation and annihilation operators as follows : @xmath93 where @xmath94 and summation runs over the three - dimensional wave - vector @xmath95 and over three polarizations of 3d phonons denoted by @xmath96 . here
@xmath97 is the polarization vector of a phonon , @xmath98 is the mass of the unit cell , @xmath99 is the number of unit cells in the sample , @xmath100 is the position vector .
sound velocities @xmath101 are much smaller than the spin - boson velocity @xmath79 : @xmath102 for a chain specified by @xmath103 the interaction hamiltonian , eq .
( [ hsp ] ) , can be written , using eqs .
( [ u ] ) and ( [ phi ] ) , as : @xmath104 with @xmath105 where @xmath32 and @xmath106 are the 1d momenta of spin bosons and @xmath95 is the 3d momentum of a phonon .
note that only component of the total momentum along the chain is conserved , which is explicitly given by the @xmath107-symbol .
the projection of the polarization vector on the @xmath108-axis @xmath109 is equal to : @xmath110 where two answers for the transverse phonons correspond to two possible choices of polarization @xmath111 .
it is convenient to choose the first of these polarizations to lie in the plane given by @xmath112 and @xmath95 and the second one to be normal to this plane .
since the projection @xmath113 is zero in the second case , such phonons do not couple to the spin bosons and are not discussed in this paper any further .
altogether , the hamiltonian ( [ hint ] ) describes emission and absorption of phonons by the spin - density excitations represented by the spin bosons , see fig .
[ phonon_diagrams ] . _ spin - impurity hamiltonian . _ in addition to interacting with phonons spin bosons are also scattered on impurities .
microscopically , the origin of that effect is simple : impurity leads to a local variation of the superexchange coupling @xmath2 , which leads to scattering of magnetic excitations .
the impurities affect the low temperature transport properties of the tomonaga - luttinger liquid in a dramatic way because they act as almost ideal backward scatterers for the low - lying excitations .
the most important ( relevant in rg sense ) part of the impurity hamiltonian for an impurity located at @xmath114 can be written as : @xmath115 although the effect of impurities is very strong for the low - lying excitations , for small @xmath116 and not too low temperatures the impurity scattering can be considered perturbatively .
_ boltzmann equation . _ in order to study the relaxation of spin excitations on phonons we will solve the boltzmann equation for spin bosons coupled to the bath of 3d phonons . the stationary boltzmann equation has the form : @xmath117 , \label{be}\end{aligned}\ ] ] where @xmath118 is the spin - boson distribution function and @xmath119 $ ] is the collision integral . in general , the collision integral is a non - linear functional of both @xmath118 and the phonon distribution function @xmath120 : @xmath121 $ ] .
however , as we argued in the introduction , one can assume that the relaxation of phonons is quicker than that of the spin bosons .
thus , even in the presence of the temperature gradient @xmath122 , phonons will be treated as if they are in a local equilibrium with themselves and this equilibrium is characterized by a local temperature @xmath123 : @xmath124 therefore , one can write @xmath125 = s_k [ f]$ ] . _ collision integral . _
the collision integral for spin bosons can be generally written as : @xmath126\ , \end{aligned}\ ] ] where @xmath127 stands for @xmath128 and @xmath129 is the total probability of the spin excitation to be scattered from the state @xmath32 to the state @xmath106 .
for the processes of scattering due to phonons such probabilities are given by : @xmath130 where the symbol @xmath131 stands for @xmath132 .
similar expression for the probability @xmath133 of scattering from @xmath106 to @xmath32 can be obtained by permutation of @xmath32 and @xmath106 in ( [ wkk ] ) . here
the `` elementary '' scattering probabilities @xmath134 of the spin boson due to emission or absorption of the phonon with the momentum @xmath95 and polarization @xmath96 are determined from eqs .
( [ hint ] ) , ( [ vint ] ) : @xmath135 where @xmath136 is elementary cells volume , @xmath137 is the spin - boson energy . in all the collisions the total energy and the projection of the momentum along the chain is preserved .
these conservation laws are enforced by the @xmath107-functions in eq .
( [ w ] ) . _ linearized boltzmann equation .
_ boltzmann equation ( [ be ] ) should be solved in the presence of a non - zero temperature gradient @xmath138 .
assuming that the gradient @xmath139 is small one can introduce a usual ansatz for the distribution function @xmath118 to linearize the boltzmann equation : @xmath140 where @xmath141 is a non - equilibrium correction to the equilibrium distribution function @xmath142 .
function @xmath143 is considered to be small .
since the collision integral vanishes identically in an equilibrium state when @xmath144 and @xmath145 , one expects that @xmath141 should be proportional to the temperature gradient @xmath122 .
thus , the boltzmann equation for the spin bosons to the first order in @xmath122 can be rewritten as : @xmath146 where @xmath147 now stands for the linearized collision integral which can be expressed using the notations of eq .
( [ f ] ) as : @xmath148 \ , \nonumber\\ \label{s2 } s^{(2)}_{k\ell } & = & -\int_{k'{\bf p } } w^\ell_{k'k{\bf p } } \bigg [ \left ( n_{{\bf p}\ell}^0 + f_{k}^0 + 1\right ) f_{k'}^1 \\ & & \phantom{-\int_{k'{\bf p } } w_{k'k\ell } \bigg [ } + \left(f_{k'}^0 -n_{{\bf p}\ell}^0 \right ) f_{k}^1 \bigg ] \nonumber \ .\end{aligned}\ ] ] the integral @xmath149 accounts for two types of collision events : _ ( i ) _ spin boson with the momentum @xmath32 emits a phonon with the momentum @xmath95 and scatters into the state with the momentum @xmath106 ; _ ( ii ) _ process inverse to _
( i)_. likewise , @xmath150 describes the absorption of a phonon by spin boson with the momentum @xmath32 and the corresponding inverse process , see fig .
[ phonon_diagrams ] . and ( b ) @xmath150 .
solid lines are spin bosons , wavy lines are phonons.@xcite , width=302 ] once the boltzmann equation ( [ boltzmann ] ) is solved for @xmath151 the spin - boson thermal current density @xmath152 can be written as : @xmath153 where the coefficient @xmath46 is the spin - boson thermal conductivity . the total thermal conductivity observed in experiment is the sum of @xmath46 and the phonon thermal conductivity . _
kinematic considerations .
_ to solve the boltzmann equation ( [ boltzmann ] ) one needs to evaluate integrals in eqs .
( [ s1 ] ) and ( [ s2 ] ) .
part of this task can be accomplished without any approximations as the integral over @xmath95 in @xmath149 and @xmath150 can be calculated explicitly .
because of the conservation laws discussed above the remaining integration over @xmath106 will be restricted to some finite intervals whose extent and range depend on @xmath32 .
mathematical details of these calculations are given in appendix [ app_a ] .
the collision integral derivation can be simplified even more if we restrict our attention to the spin bosons whose energies are small : @xmath154 .
these energies are the most important since the transport is dominated by the bosons with small @xmath32 .
such an approximation is used at various stages of @xmath147 calculation .
the consistency of our result with this latter assumption can be verified afterward . _ relaxation time .
_ one can write the collision integral in eq .
( [ coll_int ] ) as a sum of two terms : @xmath155 where the first term has the usual relaxation time form , while the second one does not .
however , the second term can be estimated and it is shown to be small for small spin - boson energy @xmath156 ( for details see appendix [ app_a ] ) . finally , with the help of these approximations , the collision integral can be written in the relaxation time form @xmath157 , where @xmath158 is the transport relaxation time .
since the transport is dominated by the spin bosons with small momentum , one can use that smallness to simplify significantly the corresponding expression ( [ tau ] ) in the collision integral to obtain : @xmath159 where @xmath160 is a constant proportional to the spin - lattice coupling , @xmath161 comes from the bosonic distribution function , and powers of @xmath32 and @xmath106 are from the scattering amplitudes @xmath162 in eq .
( [ vint ] ) and due to the integration over @xmath95 .
what does this formula tell us about the most effective scattering process for a spin boson with small momentum ?
one can see that the major contribution to the integral in ( [ tau_tr00 ] ) comes from @xmath163 .
that is , the scattered spin boson is `` thermalized '' .
therefore , the most important scattering process is the absorption of a thermalized phonon .
further simplification of the problem can be achieved by exploiting the small parameter @xmath164 . because the spin bosons are fast
, the energy and momentum conservation dictates that the majority of the phonons which interact with the spin subsystem must have their momentum almost normal to the chain direction .
indeed , the energy conservation gives that the phonon momentum @xmath165 , where @xmath32 and @xmath106 are the 1d spin - boson momenta , while the momentum conservation along the chain gives @xmath166 .
therefore , the momentum component perpendicular to the chain for a typical phonon is : @xmath167 , except for the case of almost elastic backward scattering @xmath168 .
the latter scattering event does not contribute substantially to the transport relaxation time ( see appendix [ app_a ] ) . since the momentum @xmath95 is almost normal to the chain the corresponding projection of the polarization vector for the longitudinal phonons given in eq .
( [ xi ] ) is small : @xmath169 . at the same time the transverse phonon polarization projection is @xmath170
thus , the most effective spin - phonon scattering is due to the transverse phonons .
we would like to rewrite eq .
( [ tau_tr00 ] ) in the following form : @xmath171 with the auxiliary function @xmath172 defined as : @xmath173 and @xmath38 given by : @xmath174 @xmath175 where @xmath176 is the zeta - function .
one can easily verify that : @xmath177 where @xmath178 plays the role of a crossover temperature .
that is , for the temperatures well below the debye energy @xmath179 the integration in eq .
( [ tau_tr000 ] ) is restricted by @xmath180 and the relaxation rate is given by : @xmath181 however , for the temperatures comparable to @xmath3 and higher ( in fact , for @xmath182 ) the integration limit in eq .
( [ gamma0 ] ) becomes small .
this leads to the change in the temperature dependence of the relaxation rate from @xmath183 to @xmath184 : @xmath185 the relaxation rate for the spin boson due to scattering on phonons in the form given in eq .
( [ tau_tr000 ] ) with the limiting cases given by eqs .
( [ tau_tr0 ] ) and ( [ tau_tr0_a ] ) are the main results of this section .
it is instructive to inspect and compare our results with the well - known phonon - phonon scattering rate for the 3d phonons , studied in detail many years ago.@xcite for the 3d phonon - phonon scattering rate one finds that it depends on the higher power of the wave - vector @xmath42 , while its temperature dependence is the same .
this is because the scattering space in phonon - phonon case is restricted by the momentum conservation in all three dimensions . in our case , it is only the momentum along the chains which is conserved in the spin - phonon scattering .
this leads to fewer restrictions and a larger scattering space for phonons and makes the relaxation rate @xmath41 .
thus , the characteristic power of @xmath40 in the relaxation rate eqs .
( [ tau_tr00])-([tau_tr0_a ] ) can be traced back to the fact that spin excitations are confined to lower dimensions than phonons .
this implies that our results for the spin - phonon relaxation rate should be applicable to other low - dimensional spin systems as well .
the temperature dependence of the relaxation rate is due to the bosonic nature of both the spin excitations and the phonons . _
infrared divergence .
_ with the collision integral in the relaxation time form the boltzmann equation is trivially solved : @xmath186 where we used that @xmath187 in the small @xmath32 limit .
once @xmath151 is found the energy current density can be determined from eq .
( [ j_e ] ) which yields the thermal conductivity : @xmath188 this expression diverges at small @xmath32 giving rise to an infinite @xmath46 .
the divergence happens because the scattering of spin bosons on phonons is not sufficiently strong for @xmath189 to ensure the convergence of the low - energy contribution to the thermal conductivity .
such a situation is familiar from the phonon thermal conductivity in 3d insulators.@xcite in the 3d phonon problem the regularization of such a divergence is non - trivial and involves consideration of the higher - order phonon scattering processes,@xcite degeneracy of different phonon branches at high symmetry points,@xcite and interface boundary scattering . we would like to note here that such a divergence is not related to any conservation law or integrability of the problem .
the scattering simply becomes too weak at low energies and is unable to equilibrate the excitations .
a model example of such a behavior would be a system of free , noninteracting phonons in a continuum , which are scattered only by the point - like impurities according to the rayleigh s law.@xcite all the conservation laws are broken in this case .
yet , the heat conductivity is infinite in such a system since the rayleigh scattering is ineffective at low energies , which leads to a diverging integral similar to our eq .
( [ k_diver ] ) . in our case ,
because of the 1d nature of spin system the impurity scattering of spin bosons is very effective at low energies , as we will show in the next section , so that even weak impurity potential renders the thermal conductivity finite .
in a 1d system impurities have a dramatic effect on the low - energy excitations .
the hamiltonian for spin excitations in a 1d spin chain interacting with a single defect is given in eq .
( [ himp ] ) . since the impurity hamiltonian is a relevant perturbation in the rg sense , defects scatters the low - energy excitations very effectively destroying their quasi - ballistic propagation .
therefore , the disorder in the magnetic coupling @xmath2 will remove the infrared divergence in eq .
( [ k_diver ] ) .
the impurity hamiltonian in eq .
( [ himp ] ) is not a low - order polynomial of the bosonic field .
thus , it is impossible to account for the impurity scattering within the formalism of the boltzmann transport theory
. instead , we will evaluate the spin boson life - time @xmath190 using the green s function perturbative expansion in powers of @xmath191 .
there are two issues we must clarify before proceeding with this approach .
first , below the kane - fisher temperature the impurity scattering is very strong and leads to the localization of 1d excitations.@xcite on the other hand , at the temperatures well above this scale the impurity contribution can be analyzed perturbatively .
thus , for the perturbation theory in powers of @xmath191 to be valid the temperature must be bigger than the kane - fisher temperature @xcite : @xmath192 second , it is important to note that , generally , the life - time of an excitation is not equivalent to the transport relaxation time . while the only processes which violate the conservation of the momentum contribute to the latter , any kind of scattering shortens the former .
however , since eq .
( [ himp ] ) does not conserve momentum and since in any typical scattering event the spin - boson momentum changes drastically , there is no distinction between these two time scales in our case .
this justifies the use of our approach .
the correction to the single - boson green s function from the impurity scattering is evaluated as follows .
the lowest - order single impurity contribution to the green s function is : @xmath193 the right hand side of this equation is proportional to the self - energy .
since the impurity breaks the translational invariance the self - energy depends on two momenta . as usual , the translational invariance will be restored after averaging over the random impurity positions . since the perturbation @xmath194 ,
eq . ( [ himp ] ) , is an exponential in the bosonic field @xmath195 rather than a polynomial , the calculation of the matsubara average involves an effective summation of an infinite number of terms , as schematically shown in fig .
[ imp_scatt ] .
details of such a procedure are given in appendix [ app_b ] .
as a result of our calculation the retarded self - energy is obtained though the analytical continuation of the matsubara self - energy : @xmath196 with @xmath50 being the dimensionless impurity concentration .
note that this self - energy is pure imaginary . and @xmath197 .
shaded ellipses emphasize that there is a summation over virtual states with any number of spin - boson excitations .
this is due to the fact that the interaction hamiltonian is exponential in the bosonic field .
formally , the ellipses correspond to @xmath198.@xcite , width=302 ] once the self - energy is found one can obtain the dressed green s function : @xmath199 where @xmath200 from here it is possible to find the lifetime of the spin bosons by solving equations on the poles of the green s function : @xmath201 .
the yields the following expression for @xmath202 : @xmath203 where @xmath204 the details of the derivation are given in appendix b. we would like to note that the impurity - induced relaxation rate @xmath205 has been first obtained by oshikawa and affleck , ref .
, in the context of the theory of electron spin resonance in the s=1/2 quantum antiferromagnetic chains .
unlike the spin - phonon scattering , the relaxation rate due to the disorder is independent of @xmath32 .
the impurity scattering provides an effective relaxation mechanism at low energies and thus removes the divergence of the thermal conductivity in eq .
( [ k_diver ] ) by effectively cutting off the low - energy spin bosons .
now we can start evaluating the self - energy corrections to the spin boson green s function .
our first step is the calculation of the matsubara green s function @xmath522 . since the perturbation is an exponential of the bosonic field @xmath523 rather than a polynomial we present our calculation in details .
the second order correction to the green s function is given by : @xmath524 the triangular brackets stand for the matsubara time - ordered averaging , and @xmath525 is the impurity position .
the bare boson green s function @xmath526 is a fourier transform of @xmath527 . to calculate the expression in the triangular brackets in eq .
( [ d2 ] ) we must expand both exponents into taylor series @xmath528 and apply wick s theorem . the external operators @xmath529 can be contracted with fields @xmath523 in four possible ways schematically shown in fig .
[ imp_scatt ] : ( a ) both external operators contracted with the monomial @xmath530 , ( b ) both external operators contracted with the monomial @xmath531 , and ( c ) and ( d ) one external operator contracts with @xmath530 while another contracts with @xmath531 .
each contraction with the external operator acts on a monomial as a derivative with respect to the field @xmath523 times @xmath532 or @xmath533 .
effectively , contraction with an external operator is equivalent to differentiation with respect to @xmath523 times some @xmath534-number .
because of that one can derive : @xmath535 the first term in the curly brackets corresponds to the diagram ( a ) in fig .
[ imp_scatt ] , the second corresponds to the diagram ( b ) and so on .
the gray bubbles in fig .
[ imp_scatt ] correspond to @xmath536 . as we see from ( [ aux ] ) , the impurity scattering explicitly violates the momentum conservation .
it will be restored once we average over the impurity positions .
such impurity averaging leads to a simple change in the above expression : @xmath537 , where @xmath50 is the dimensionless impurity concentration .
after that the impurity position @xmath525 is arbitrary and can be put to zero .
this yields : @xmath538 with the help of the standard formula for free bosonic field : @xmath539 the above equation can be rewritten in the @xmath540-space as : @xmath541 the average of the bosonic fields can be evaluated as follows @xmath542 it is possible to evaluate the integral @xmath543 for small ( @xmath544 ) and large ( @xmath545 ) matsubara frequency @xmath540 : @xmath546 the fact that at large @xmath547 function @xmath543 decays as an inverse @xmath548 guarantees the convergence of the matsubara sum .
unfortunately , it is impossible to continue our derivation analytically due to complex properties of @xmath549 . to overcome this difficulty we utilize a trick . rather than using the actual form of @xmath543 we put a different function in its place : @xmath550
such a replacement preserves @xmath551 properties of the sum but radically improves convergence at large @xmath540 .
since we are not interested in the high energy ( @xmath552 ) , short time ( @xmath553 ) properties of @xmath554 this procedure is justified .
now the matsubara summation can be done and it leads to : @xmath555 in deriving this formula we used the identity : @xmath556 which is correct for any complex @xmath557 , @xmath558 . to prove this formula one can expand the logarithm in taylor series with respect to powers of @xmath559 .
once @xmath560 is calculated it must be substituted into ( [ sigma ] ) .
now we have to evaluate the fourier integral : @xmath561 this integral can be transformed into an integral over a unit circle in the complex plane : @xmath562 the polynomial under the square root has three zeros : @xmath563 and @xmath564 .
our branch - cut consists of two components : the first component stretches from @xmath565 to @xmath566 , the second component stretches from @xmath567 to @xmath568 .
such a choice of the branch - cut ensures that the integration path , the unit circle , does not cross the branch - cut , see fig .
[ contour ] . ) is shown .
the thick gray lines represent the branch cuts .
the unit circle is transformed into a new contour circumventing one of the branch cuts .
this new contour is shown by the dash line.,width=226 ] it is possible now to change the integration path .
the new integration path runs counterclockwise along the sides of @xmath569 component of the branch - cut . in turn , this integral can be transformed into an integral over the real axis : @xmath570 finally , the matsubara self - energy equals to @xmath571 to obtain the retarded self - energy @xmath572 the analytical continuation @xmath573 must be done . for @xmath574
it gives : @xmath575 once the self - energy is calculated one can find the retarded green s function : @xmath576 and extract the lifetime from the position of the green s function pole : @xmath577 as a result the lifetime @xmath578 is given by : @xmath579 with the auxiliary parameter @xmath245 is defined as @xmath580 where @xmath50 is the dimensionless impurity concentration .
j. takeya , i. tsukada , y. ando , t. masuda , k. uchinokura , i. tanaka , r. s. feigelson , and a. kapitulnik , phys .
b * 63 * , 214407 ( 2001 ) ; j. takeya , i. tsukada , y. ando , t. masuda , and k. uchinokura , phys .
b * 62 * , r9260 ( 2000 ) ; a.m. vasilev , v.v .
pryadun , d.i .
khomskii , g. dhalenne , a. revcolevschi , m. isobe , and y. ueda , phys .
. lett . * 81 * , 1949 ( 1998 ) .
r. l. douglass , phys .
rev . * 129 * , 1132 ( 1963 ) ; d. c. mccollum , r. l. wild , and j. callaway , phys . rev . * 136 * , a426 ( 1964 ) ; d. walton , j. e. rives , and q. khalid , phys .
b * 8 * , 1210 ( 1973 ) ; d. douthett and s. a. fiedberg , phys . rev . *
121 * , 1662 ( 1961 ) .
_ cumulative result .
_ finally , having at hands the transport relaxation rates of the spin - boson due to scattering on the phonons , eq .
( [ tau_tr000 ] ) , and on impurities , eq .
( [ tau_imp ] ) , we are adequately equipped to calculate the thermal conductivity of the spin chains . the total relaxation time is given by : @xmath209 then the solution of the boltzmann equation which we had for the phonon scattering ( [ f1_ph ] ) should be modified to : @xmath210 as a result the thermal current is given by : @xmath211 thus , using the explicit expressions for the spin - phonon and spin - impurity relaxation rates the following formula for the thermal conductivity can be obtained : @xmath212 where @xmath213 and @xmath214 with @xmath172 defined in eq .
( [ gamma0 ] ) .
since the temperature is always much smaller than @xmath2 one can safely replace the upper limit in eq .
( [ kappa2 ] ) by infinity .
( [ kappa2 ] ) specifies a function with the following properties .
depending on the value of @xmath3 it either has a single maximum at some @xmath215 ( larger @xmath216 ) or is a monotonic function ( smaller @xmath216 ) . in both cases this function vanishes as @xmath217 at @xmath218 and saturates at @xmath62 .
these qualitative features of @xmath59 can be easily established . without impurities ( @xmath219 ) the integral in eq .
( [ kappa2 ] ) diverges at the lower limit .
the impurity scattering cuts off this divergence .
when the temperature is small the impurity scattering is much stronger than the typical spin - phonon scattering , which corresponds to @xmath220 .
thus , @xmath4 at low temperatures . in the limit of larger temperatures
the impurity scattering is much weaker than the spin - phonon one and @xmath221 .
if we assume that the temperature is still much less than the debye energy @xmath222 then @xmath223 . therefore , in this temperature range eq .
( [ kappa2 ] ) yields @xmath224 . at
yet higher temperatures @xmath225 the temperature behavior of the spin - phonon relaxation rate changes from @xmath183 to @xmath184 , which is reflected in the change of @xmath226 from 1 to @xmath227 , see eqs .
( [ tau_tr000])-([tau_tr0_a ] ) .
this gives @xmath228 .
altogether , assuming @xmath229 is not too small : @xmath230 where @xmath231 , as introduced in sec . [ phonons ] . note that the low - temperature , impurity - controlled thermal conductivity in the generic luttinger liquids has been studied in ref . using the memory - matrix formalism .
our @xmath232 result matches the result of that work for the luttinger liquid parameter of the spin-@xmath64 chain @xmath233 . as one
can see from eq .
( [ kappa_q1 ] ) our results do not support suggestion of @xmath234 made in the experimental work , ref . .
below we will discuss the origin of the maximum and of the saturation value of @xmath46 , but first we compare with the available experimental data . _ comparison with experiments . _ to compare our results with experimental data it is convenient to rewrite the above expression for @xmath59 , eq .
( [ kappa2 ] ) , using the `` reduced '' temperature units @xmath235 with @xmath58 which corresponds to the maximum in the thermal conductivity : @xmath236 where @xmath237 from eq .
( [ gamma0 ] ) , and the constants @xmath238 and @xmath239 are chosen in such a way that the temperature @xmath240 indeed corresponds to a maximum in @xmath46 and that @xmath241 is indeed equal to @xmath242 . for @xmath229 not too close to @xmath58 the constants are : @xmath243 and @xmath244 .
such a choice of variables simply replaces a combination of impurity- and lattice - related coupling constants encoded in our @xmath245 and @xmath38 , whose actual values are generally unknown , by phenomenological constants @xmath58 and @xmath242 , found from experiment .
in fact , in the limit @xmath246 this procedure completely determines our @xmath247 given in eq .
( [ kappa_u ] ) since there are no adjustable parameters left .
the resulting @xmath59 for this limiting case is shown in fig .
[ chain ] by the dashed line .
normalized to its maximal value vs reduced temperature @xmath248 .
diamonds are the experimental data for sr@xmath0cuo@xmath1 , ref . .
@xmath249k and @xmath250wm@xmath251k@xmath251 for this material .
shaded area schematically represents the range where the phonon background subtraction creates a large uncertainty in the data.@xcite solid line is the results of this work , eq .
( [ kappa_u ] ) , for @xmath252 .
arrows mark the saturation value @xmath253 for the solid line at @xmath254 and the spin - phonon scattering crossover scale @xmath255 .
dashed line corresponds to @xmath59 , eq .
( [ kappa_u ] ) , for @xmath256 .
[ these results were recently presented in ref .
.],width=321 ] once we changed the variables to eq .
( [ kappa_u ] ) there is only one parameter to adjust : @xmath3 .
if we choose @xmath257 we fit experimental data for the spin - chain compound sr@xmath0cuo@xmath1 perfectly . for this material @xmath258 and @xmath250wm@xmath251k@xmath251 .
figure [ chain ] shows the comparison of our results and experimental data , where the latter are shown by diamonds and the former is the solid line .
shaded area highlights the region where , experimentally , the phonon background subtraction creates a large uncertainty in the data.@xcite note that at high temperatures @xmath259 eq .
( [ kappa_u ] ) yields the asymptotic value of the spin - boson thermal conductivity @xmath253 : @xmath260 therefore , when we fix @xmath3 we fix @xmath261 as well .
our choice of @xmath3 corresponds to @xmath262 . in fig .
[ chain ] @xmath263 is marked by an arrow .
we would like to emphasize that apart from @xmath264 and @xmath58 there is only one adjustable parameter , @xmath3 , which is used in our fitting procedure , and yet our theory yields an excellent agreement with the experimental data over the whole temperature range .
note that the transition between intermediate- and high - temperature regimes for @xmath46 is determined by the scale @xmath229 , which separates different temperature behavior for the spin - phonon scattering as discussed in sec .
[ phonons ] .
it is this scale , not the debye temperature itself , is a parameter of our theory . for sr@xmath0cuo@xmath1
this scale is found to be @xmath265k , shown in fig .
[ chain ] by an arrow .
the value of the debye temperature @xmath3 that comes out from our fit of thermal conductivity for sr@xmath0cuo@xmath1 is @xmath266k , which is higher than the estimations of @xmath3 for this material obtained by a different method.@xcite since in our theory the debye temperature is a characteristic energy scale for the phonon bandwidth this may imply that the spin system is also coupled to the optical phonons .
in fact , the typical width of the whole phonon spectra in the cuprates is about 900k.@xcite to complete our comparison we provide here the asymptotic expressions for @xmath59 in the low- and intermediate - temperature regimes . at small temperature @xmath267 the thermal conductivity is : @xmath268 in this regime the conductivity is controlled entirely by impurities .
when the temperature is higher the number of high - energy spin bosons increases .
the most important relaxation mechanism for these excitations is the spin - phonon scattering .
however , the low - energy spin bosons are still scattered mostly by impurities .
thus , at intermediate temperature @xmath269 both mechanisms contribute to the relaxation of the thermal current : @xmath270 these asymptotic formulas can be tested experimentally , although the intermediate - temperature regime for sr@xmath0cuo@xmath1 is not very well pronounced because @xmath229 is only 3 times larger than @xmath58 .
such an intermediate asymptotic behavior might be more relevant for some other systems where the difference between @xmath58 and @xmath229 is larger . for larger temperatures @xmath271 the thermal conductivity saturates .
thus , one of the most straightforward ways to verify our theory would be to measure the thermal conductivity for higher temperatures and check if the saturation really takes place as we predict .
_ different choices of @xmath3 .
_ one may wonder at this point how much variation in the temperature dependence of @xmath46 can be expected from the theory given by eq .
( [ kappa2 ] ) or ( [ kappa_u ] ) if the parameters of the model , such as @xmath3 , are varied .
as is mentioned above there are three independent parameters : phonon- and impurity - couplings @xmath38 and @xmath245 , and the debye temperature @xmath3 . to generate a sequence of curves for the theoretical @xmath46 v.s .
@xmath28 the following procedure is convenient .
since at @xmath276 thermal conductivity given by eq .
( [ kappa2 ] ) saturates at a constant value , it is convenient to normalize @xmath59 to its value at @xmath277 , i.e. @xmath253 , not to @xmath242 .
this is also because for @xmath278 @xmath59 becomes a monotonic function and the determination of @xmath242 is ambiguous .
such a normalization reduces the number of free parameters to two .
it is natural to choose one of those parameters to be a phonon crossover scale @xmath216 ( @xmath279 ) and to define the other one as a reference temperature @xmath10 . using eq .
( [ kappa_u ] ) with dimensionless constants @xmath238 and
@xmath239 related through : @xmath280 one can generate a sequence of curves for @xmath281 v.s .
@xmath282 for various values of @xmath274 shown in fig .
[ chain_theta ] .
results for @xmath2835 , 4 , 3 , 2 , 1.33 , and 1 are shown ( @xmath275 decreases from the top to the bottom curve ) . the value of @xmath244 , same as in fig .
[ chain ] , is used to ensure that for @xmath284 ( in reality , for @xmath285 ) the reference temperature automatically corresponds to the temperature of the maximum , i.e. @xmath286 .
different choice of @xmath239 would simply correspond to rescaling of @xmath10 leaving the form of @xmath247 unchanged .
thus , fig .
[ chain_theta ] gives a description of possible variations in the results of the theory given by eq .
( [ kappa2 ] ) within the experimentally relevant range of @xmath287 . as is described above , in order to fix the remaining two parameters and fit the experimental data
one may associate the experimental value of @xmath58 with the position of the theoretical maximum and then vary @xmath216 to match @xmath288 , for example . a similar strategy has been used to obtain fig .
[ chain ] where the best fit closely corresponds to @xmath289 curve in fig .
[ chain_theta ] . if the experimental curve does not have a maximum,@xcite one can choose any other reasonably distinct @xmath290 in order to fix the choice of theoretical parameters @xmath10 and @xmath216 .
_ origin of @xmath58 .
_ the temperature @xmath58 at which @xmath46 reaches its maximum can be expressed through the original parameters of the problem as : @xmath291 in a similar fashion one can relate @xmath242 to the microscopic parameters as : @xmath292 the physical meaning of @xmath58 is the following .
it is possible to define the `` thermal '' spin - phonon relaxation rate : @xmath293 where @xmath294 . by its definition @xmath295
is independent of @xmath32 .
it characterizes rate of the spin - phonon collisions of the thermal spin bosons at a given temperature .
the scale @xmath58 emerges naturally as a solution of the equation : @xmath296 in other words , when @xmath297 the thermal spin - phonon and spin - impurity collision rates are the same . at the temperature higher ( lower ) that @xmath58 the spin - phonon ( impurity ) scattering dominates .
note that in the low - temperature regime we assume that the temperature is still much larger than the kane - fisher temperature @xmath47 .
_ origin of @xmath253 .
_ our work predicts a remarkable behavior of the thermal conductivity at the temperatures higher than the scale related to the debye energy : saturation at a constant value , eq .
( [ kappa_as ] ) .
a parallel can be drawn with the thermal conductivity of metals , where the electronic part of @xmath298 saturates at the temperatures above the debye energy.@xcite the parallel can be suspected to be even deeper because of the presence of the large energy scales in both problems , fermi energy @xmath299 in the case of metals and @xmath300 in our case .
a closer examination , however , reveals a difference between the two cases . in metals ,
the thermal conductivity temperature behavior is defined by the electronic specific heat @xmath301 and the electron - phonon relaxation time @xmath302 through the quasiclassical relation : @xmath303 .
note that it is the thermalized electrons which contribute most substantially to the transport .
specific heat is linear in @xmath28 , while at @xmath37 the relaxation time is inversely proportional to the temperature due to the thermal population of phonons , @xmath304 .
altogether , this renders @xmath305 . in our case
the excitations are spin bosons , not fermions at large momentum @xmath306 .
therefore , in contrast , in the phonon - scattering dominated regime @xmath307 the spin - boson relaxation rate is strongly @xmath32-dependent and the major contribution to the thermal current comes from the long - wavelength bosons with @xmath308 .
thus , the impurity - scattering continues to contribute to the temperature dependence of the thermal conductivity at higher temperatures .
therefore , in the case of spin chains the saturation of @xmath46 is a result of a non - trivial combination of : particular temperature- and @xmath32-dependencies of the impurity and phonon relaxation rates and the 1d density of states of the spin - bosons .
_ further predictions .
_ our equations ( [ kappa_as ] ) , ( [ tm ] ) and ( [ kmax ] ) allow us to formulate several predictions of our model for the impurity concentration dependence of several quantities that can be verified experimentally .
first , we see that the temperature @xmath309 at which @xmath59 reaches its maximum scales as @xmath310 .
such a weak dependence means that for a particular material the maximum in @xmath46 should be around the same temperature for a broad range of disorder concentration .
second , the maximum value of the thermal conductivity @xmath311 scales as @xmath312 , eq .
( [ kmax ] ) .
third , the saturation value @xmath253 has yet different concentration dependence @xmath313 .
such behaviors can be looked for in the materials with the isotope substitution for either magnetic ions ( cu@xmath314 ) or surrounding ions as they all will lead to the local modification of the superexchange constant .
if such dependencies are observed this will provide strong support to our theory .
another interesting experimental suggestion is based on the specifics of the spin - phonon scattering which we have found in sec . [ phonons ] .
namely , the most important scattering of spin bosons in the phonon - dominated regime is due to phonons whose momentum is almost normal to the direction of the spin chains .
the characteristic energy of such phonons is @xmath28 .
one can think of the following experimental setup : the heat current is directed along the chains while the thermal phonon pulse is induced perpendicular to the heat current ( e.g. , by a short laser pulse ) .
this should lead to a sizable suppression of the spin - boson thermal current due to additional scattering by the extra phonons .
the increase of the phonon heat current due to transverse pulse will be either negligible or can be accounted for by comparing to the results of the same experiment with the heat current and heat pulse both directed perpendicular to the chains .
_ consistency check .
_ in considering the spin - boson relaxation due to phonons we emphasized the importance of the two - stage , bottle - neck process , in which the spin boson scatters on the phonon via a `` normal '' process , while the momentum relaxation via an umklapp process is taken care of by the phonon bath . having carried out a comparison with experiments we can now verify the consistency of our results with the initial assumption .
namely , we can now check whether the characteristic spin - phonon relaxation time @xmath19 is indeed much longer than the characteristic phonon - phonon relaxation time @xmath22 . since below @xmath58
the scattering is impurity - dominated we need to estimate the relaxation times for @xmath315 only .
we have already introduced the `` thermal '' relaxation time for spin bosons @xmath295 in eq .
( [ tau_th ] ) . at @xmath240
one can express such relaxation time through @xmath242 , eq .
( [ kmax ] ) , as : @xmath316 the phonon relaxation time can be estimated from the quasiclassical expression for the phonon thermal conductivity : @xmath317 , where @xmath318 is the lattice specific heat and @xmath319 is the average transport relaxation time of a phonon due to phonon - phonon scattering . using these expressions we write the ratio of the thermal spin - phonon and the average phonon - phonon relaxation times as : @xmath320 where the specific heat @xmath318 can be found as @xmath321.@xcite one can see that the smallness of @xmath322 in ( [ tau_2 ] ) is canceled by the largeness of @xmath323 .
for sr@xmath0cuo@xmath1 in the region around @xmath58 the spin - boson and phonon parts of the thermal conductivity are of the same order . also , @xmath324 .
taking the actual values of @xmath242 , @xmath325 , @xmath58 , @xmath2 , etc .
for sr@xmath0cuo@xmath1 we find that : @xmath326 .
it appears , superficially , that our assumption is not valid .
however , at @xmath327 the typical momentum @xmath328 of a spin boson contributing to the thermal transport is much smaller than the `` thermal '' momentum @xmath294 : @xmath329 .
thus , the typical relaxation time is much larger than the thermal one @xmath330 .
we also recall here that the phonons contributing to our two - stage process most effectively are the thermalized phonon with the momentum @xmath331 as shown in sec . [ phonons ] . therefore , @xmath30 of such a phonon is much shorter than the average transport relaxation time @xmath332 which we estimated above , @xmath333 .
thus , the characteristic spin - phonon and phonon - phonon times involved in the two - stage process obey : @xmath334 due to both @xmath335 and @xmath336 .
this proves the validity of our approach .
we also estimated the kane - fisher temperature for sr@xmath0cuo@xmath1 from @xmath337 which follows from eqs .
( [ tkf ] ) , ( [ tm ] ) , and ( [ kmax ] ) . using experimental values for @xmath242 and @xmath58
one obtains that @xmath338 which means that our assumption for the impurity - controlled regime @xmath339 is valid unless the impurity concentration @xmath50 is less than 0.1% .
we are not aware of the actual level of disorder in sr@xmath0cuo@xmath1 , but such a high purity is unlikely .
therefore , this assumption is valid too . with the help of our asymptotic formula ( [ kmax ] )
one can estimate the spin - coupling constant using experimental values of @xmath242 and @xmath58 .
this estimate leads to an unreasonably high value of @xmath340 .
first , a large value of the spin - phonon coupling constant can be due to an exponential dependence of the superexchange on the interatomic distance , as it is generated by the virtual tunnelling of electrons .
second , as one can see from eqs .
( [ tm ] ) and ( [ kmax ] ) the expression for the spin - phonon coupling contains high powers of both @xmath2 and @xmath58 .
this creates an extra sensitivity of the result of our estimate to the unknown dimensionless constants , which our theory might be unable to capture.@xcite _ other materials .
_ other materials of the cuprate family in which the anomalous thermal transport has been observed include the zig - zag chain@xcite and spin - ladder materials,@xcite and another spin - chain material , bacu@xmath0si@xmath0o@xmath341 , with much smaller superexchange constant.@xcite the most straightforward case for applying our theory seems to be the case of srcuo@xmath0 where the spin excitation spectrum remains gapless because of the frustrated coupling between the chains .
since the inter - chain coupling is of order @xmath342 one may expect that while the low- and intermediate-@xmath28 behavior of @xmath46 should be the same as in the case of sr@xmath0cuo@xmath1 , the high-@xmath28 part might be different due to an additional scattering mechanism from the inter - chain interaction .
however , as we show in section [ meanfp ] , the high-@xmath28 regime is fully dominated by the long - wavelength spin excitations , which means that @xmath343 should remain irrelevant and the behavior of @xmath59 for srcuo@xmath0 should be very similar to the `` simple '' chain material sr@xmath0cuo@xmath1 at all temperatures .
nevertheless , we were unable to fit the experimental data for srcuo@xmath0@xcite as successfully as for sr@xmath0cuo@xmath1 . on the other hand , as we learned from a very recent work ref .
there might be an issue with the subtraction background for these materials .
the thermal conductivity of the 5% ca - doped srcuo@xmath0 does exhibit a saturation at high - temperatures , in an excellent agreement with our theory , but the peak at lower temperatures is much less pronounced .
we therefore refrain from an explicit comparison of our theory to srcuo@xmath0 until the experimental issues are settled . in the case of
( la , ca , sr)@xmath14cu@xmath15o@xmath16 materials the excitations are gapped .
therefore , the results of our work can not be straightforwardly applied to them because the gap , not the impurity scattering , controls the low - energy cut - off scale .
thus , one needs to reconsider the problem of the thermal transport for the gapped spin systems . on the other hand , the spin - phonon scattering part of such
a consideration may stay very similar to our case , especially for ca@xmath344la@xmath345cu@xmath15o@xmath16 .
this latter compound presumably contains a lot of 3d disorder ( not related to the spin subsystem ) due to a random distribution of ca and la ions,@xcite which might help to facilitate the fast phonon momentum relaxation and ensure that @xmath346 at high temperatures .
we would like to note that another very recent experimental work , ref .
, has discussed the role of the spin - phonon coupling in the gapped systems using a very different framework . the case of another spin - chain material , bacu@xmath0si@xmath0o@xmath341 ,
is also different because of the lack of the spin and lattice energy scales separation . in this compound
the spin boson velocity is of the same order as the sound velocity , @xmath347.@xcite therefore , the kinematic considerations of our work will not be applicable for the collision integral in this system . in the temperature regime @xmath348 , easily achievable for this material , the thermal transport will be influenced by various other relaxation processes , which are negligible in our case .
altogether , the differences in the underlying spin models for some of these materials will , most certainly , lead to different temperature behaviors of @xmath46 .
therefore , a detailed microscopic study in these systems along the lines of our work is necessary .
having described successfully the temperature dependence of the thermal conductivity in sr@xmath0cuo@xmath1 we would like to discuss the length scales that characterize scattering in this system .
we also would like to address the question of whether the concept of the mean free path can be applied to describe thermal conductivity by spin bosons . more specifically
, we would like to ask whether the thermal conductivity of spin bosons can be written in a simplified form ( sometimes referred to as the simple kinetic equation ) : @xmath349 and if yes in what regime . here
@xmath350 is the specific heat of the spin chains and @xmath351 is the mean free path . in the end
, we also verify that the wavelength of a typical excitation contributing to the thermal current in sr@xmath0cuo@xmath1 is shorter than its mean free path .
this condition is needed for the applicability of the boltzmann equation formalism ( quasiclassical approximation ) .
the `` total '' relaxation time of the spin boson with the wave - vector @xmath32 is defined in eq .
( [ tau_tot ] ) .
thus , one can introduce corresponding @xmath32- and @xmath28-dependent length scale : @xmath352 in order to be clear on terminology we would like to refer to this @xmath32-dependent length scale as to the `` @xmath32-dependent mean free path '' of the spin boson . at the same time the notion `` mean free path '' will be reserved for an averaged , only @xmath28-dependent , length scale . _ cut - off length .
_ spin - phonon scattering rate in eq .
( [ ell_kt ] ) depends on @xmath32 as @xmath353 while impurity scattering rate is @xmath32-independent and plays the role of a cut - off scale .
therefore , it is natural to introduce the `` cut - off '' length as : @xmath354 using explicit expression for @xmath355 , eq .
( [ tau_imp ] ) .
it is interesting to find out the absolute value of such a length for the real system like sr@xmath0cuo@xmath1 . since all the impurity- and phonon - related parameters , @xmath245 , @xmath38 , and @xmath356 , have been related to the phenomenological constants @xmath58(@xmath357 ) and @xmath242(@xmath358 w m@xmath251k@xmath251 ) , one can re - express @xmath245 through them and obtain : @xmath359 where @xmath360 and @xmath361 and @xmath362 are the lattice constants along and perpendicular to the chains in sr@xmath0cuo@xmath1 , respectively.@xcite such a large value of @xmath363 ( @xmath364 at @xmath240 ) is in agreement with the estimates for the mean free path made in the experimental works.@xcite note that the cut - off length _ grows _ as the temperature increases
this is a consequence of the impurity scattering in 1d spin - chains which is non - perturbative in the number of spin bosons .
-dependent mean free path @xmath365 v.s .
@xmath213 for three representative temperatures : @xmath240 ( solid ) , @xmath366 ( dot - dashed ) , and @xmath367 ( dashed ) .
arrows mark the cut - off length @xmath368 for each temperature .
an approximate boundary of the bosonic thermal population is marked by the shaded area.,width=302 ] _ @xmath32-dependent mean free path .
_ we now express the @xmath32-dependent mean free path ( [ ell_kt ] ) through the cut - off length ( [ ell_coff ] ) using the explicit form of the spin - phonon scattering rate , eq .
( [ tau_tr000 ] ) : @xmath369^{-1},\end{aligned}\ ] ] where @xmath370 is the auxiliary crossover function introduced in eq .
( [ gamma0 ] ) , which is related to the debye temperature : @xmath371 for @xmath372 and @xmath373 for @xmath374 . our fig .
[ ell_x ] shows the @xmath32-dependent mean free path @xmath365 v.s .
@xmath213 for three representative temperatures : @xmath240 ( solid ) , @xmath366 ( dot - dashed ) , and @xmath367 ( dashed ) . from eq .
( [ ell_xt ] ) one can see that @xmath375 is a lorentzian with the height given by the cut - off length @xmath376 ( marked by arrows in fig .
[ ell_x ] ) and a temperature - dependent width .
this width depends on @xmath28 quite strongly : @xmath377 for @xmath372 and @xmath378 for @xmath374 .
the spin boson energy normalized by to the temperature @xmath213 is a natural variable to describe thermal processes .
the values of @xmath108 which are of significance for such processes are @xmath379 because of the bosonic thermal population ; this boundary is marked by the shaded area in fig .
[ ell_x ] .
one can see that the @xmath32-dependent mean free path is essentially constant for @xmath380 in the impurity - dominated regime @xmath366 , while it depends on @xmath108 very strongly in the phonon - dominated regime @xmath367 .
_ impurity - dominated regime .
_ already from this observation one can see that in the impurity - dominated regime all the spin bosons involved in the thermal transport are characterized by approximately the same mean free path @xmath381 , fig .
[ ell_x ] dash - dotted line , and thus the mean free path concept readily applies here .
namely , one can re - write our expression for the thermal conductivity , eq .
( [ kappa2 ] ) , somewhat differently , using the @xmath32-dependent mean free path from eqs .
( [ ell_kt ] ) and ( [ ell_xt ] ) : @xmath382 since @xmath375 is @xmath108-independent in the impurity - dominated regime , this expression can be reduced to the form ( [ kinetic ] ) @xmath383 with the mean free path equal to the cut - off length @xmath384 , eq .
( [ ell_coff1 ] ) , and the specific heat given by : @xmath385 where @xmath386 and the well - known value of the integral ( @xmath387 ) have been used .
thus , our result for the mean free path in the low-@xmath28 regime ( @xmath388 ) is @xmath389 , which disagrees with experimental works where often a constant mean free path at low temperatures is reported.@xcite however , this is the regime where experimental data for the spin part of the thermal conductivity are unreliable due to large phonon background .
( @xmath184 , dashed ) and the `` thermal '' length @xmath390 ( dashed - dotted ) . in the phonon - dominated regime
@xmath391 for @xmath392 and @xmath377 for @xmath393 ) .
the `` effective '' mean free path @xmath394 , eq .
( [ ell_eff ] ) , and the approximate `` effective '' mean free path @xmath395 , eq .
( [ ell_eff_app ] ) , are also shown ( solid and dotted , respectively ) . ,
width=302 ] _ phonon - dominated regime .
_ one can see from fig .
[ ell_x ] that the situation for @xmath375 is quite different in the phonon - dominated regime . while the cut - off length grows as the temperature increases , the range of the energies of the spin bosons effectively participating in the thermal transport shrinks .
in fact , it is a very unusual feature of the problem : as the temperature increases the thermal current becomes dominated by the excitations with longer and longer wavelengths .
therefore , our approximations related to spin excitations , such as the use of bosonization language , become better justified for higher temperatures . using eq .
( [ ell_xt ] ) one can define the `` typical '' momentum of the spin boson @xmath396 , corresponding to the width of the lorentzian @xmath397 : @xmath398 where the values @xmath399 and @xmath400 for sr@xmath0cuo@xmath1 were used .
evidently , at @xmath327 one can not possibly assume that all the spin - boson excitations are characterized by the same @xmath32-dependent mean free path .
therefore , one can not reduce eq .
( [ kappa_ell ] ) for @xmath59 to a simplified form , eq .
( [ kinetic ] ) .
in other words , the mean free path concept can not be , strictly speaking , applied to the phonon - dominated regime of the problem .
this is further demonstrated in our fig .
[ ell_t ] which shows the temperature dependence of the cut - off length @xmath363 ( @xmath184 , dashed ) and the `` thermal '' length @xmath390 ( @xmath32-dependent mean free path for the excitation with the energy @xmath401 , dashed - dotted ) . the latter length , while indistinguishable from the cut - off length in the impurity - dominated regime @xmath388 , dies off very quickly in the phonon - dominated regime @xmath327 ( as @xmath402 for @xmath392 and then as @xmath377 for @xmath393 ) .
this is simply to reiterate our point that the spin bosons are characterized by strongly @xmath32-dependent mean free paths for @xmath327 .
_ `` effective '' mean free path . _
nevertheless , one may want to insist on using the simplified equation ( [ kinetic ] ) in order to extract some `` effective '' mean free path @xmath394 characterizing the system , an approach used in the experimental works,@xcite @xmath403 using the explicit form of @xmath350 obtained above ( [ c_s ] ) the `` effective '' mean free path @xmath394 can be written as : @xmath404 this integral can be taken numerically and @xmath394 is plotted in fig .
[ ell_t ] ( solid line ) .
note again that all three lengths , @xmath376 , @xmath405 , and @xmath394 are indistinguishable for @xmath388 while they are very different for @xmath327 . recalling that the height of the lorentzian @xmath375 grows @xmath184 while the width is decreasing @xmath377 for @xmath372 and @xmath378 for @xmath374 one
can immediately conclude that @xmath394 should decay as @xmath378 for @xmath372 and @xmath406 for @xmath374 as marked in fig .
[ ell_t ] . since @xmath407 , this yields @xmath408 behavior for @xmath374 .
using the fact that the width of @xmath375 is strongly @xmath28-dependent one can obtain an approximate expression for @xmath394 in eq .
( [ ell_eff ] ) which is valid in both asymptotic limits @xmath409 and @xmath307 : @xmath410 where @xmath235 and constant @xmath411 .
it is shown in fig .
[ ell_t ] ( dotted line ) . using explicit form of @xmath412
one can obtain all the correct asymptotic behaviors of @xmath394 from it . ) , compared to the experimental results of refs . .
asymptotic fits @xmath413 for @xmath388 and @xmath414 for @xmath327 are shown by the dashed lines .
shaded area schematically represents the experimental error bars due to the phonon subtraction .
, width=302 ] our fig .
[ ell_t1 ] shows our `` effective '' mean free path , eq .
( [ ell_eff ] ) , compared to the same quantity derived in refs . from the experimentally measured @xmath59 using eq .
( [ kappa_eff ] ) .
asymptotic fits @xmath413 for @xmath388 and @xmath414 for @xmath327 are also shown .
shaded area schematically represents the experimental error bars due to the phonon subtraction .
_ saturation in @xmath46 and parallel with metals .
_ we would like to return to the discussion of the origin of saturation in @xmath59 at @xmath393 .
one may be surprised , once again , by the seemingly identical features of the high-@xmath28 thermal conductivity in both the spin chains and metals : ( i ) in both cases there is large scale ( @xmath415 and @xmath2 , respectively ) which yields `` fast '' excitations , ( ii ) in both cases specific heat @xmath184 up to very high temperatures , ( iii ) for the saturation regime the dominant scattering is due to phonons in both cases , ( iv ) mean free path for metals and `` effective '' mean free path for spin bosons are @xmath416 in this regime .
it is very tempting to conclude that in the case of the spin chains the @xmath417 behavior in the `` effective '' mean free path is due to phonon population @xmath418 as in the case of metals .
however , it is much more subtle in this case . as we discussed above and in previous section , thermal transport in spin chains
relies on the long - wavelength bosons , even more so for the high temperatures ( see fig .
[ ell_x ] ) , and the @xmath28-dependence of the impurity scattering remains important .
simply put : if the impurity scattering would provide @xmath363 v.s .
@xmath28 different from @xmath184 , the @xmath28-dependence of the @xmath419 would be different from @xmath416 and there would be no saturation in @xmath59 .
in fact , this seems to be the case for the related frustrated ladder ( zig - zag chain ) material srcuo@xmath0 , which has gapless excitations as in spin chains , but its thermal conductivity does not seem to saturate at higher @xmath28.@xcite once again , it is an interesting and unusual situation in spin chains : it is always the long - wavelength range of excitations which is crucial for the thermal transport , temperature increase only enhances the phonon scattering and shrinks the relevant energy region for spin bosons .
thus , our results are universal and do not depend on the manner in which bosonization or other approximations break down at higher energies .
_ quasiclassical approximation .
_ it is instructive to verify the validity of the quasiclassical approximation which is implicitly used when utilizing boltzmann equation for the lower - order scatterings . one needs to show for the `` typical '' spin boson that its wavelength @xmath420 is shorter than its @xmath32-dependent mean free path @xmath421 . _
@xmath327 case .
_ consider @xmath327 first .
we have already defined the `` typical '' momentum for this case in eq .
( [ k_typ ] ) . using @xmath399 and @xmath400
specific for sr@xmath0cuo@xmath1 we find ( neglecting numerical coefficients of order one ) : @xmath422 the `` typical '' @xmath32-dependent mean free path is defined as : @xmath423 therefore , @xmath424 is always fulfilled for the phonon - dominated regime .
_ @xmath388 case .
_ in the impurity - dominated regime `` typical '' spin boson has an energy @xmath425 .
thus , the `` typical '' wavelength is : @xmath426 the `` typical '' @xmath32-dependent mean free path is simply a cut - off length : @xmath427 thus , @xmath428 at @xmath429 and the quasiclassical approximation is valid above this temperature .
the main assumption is the smallness of the phonon relaxation time as compared to the spin - phonon scattering time .
this is justified since the spin - lattice coupling is generically smaller than the phonon - phonon coupling .
this also allows us to restrict ourselves with the lowest order in the spin - phonon coupling within the boltzmann equation formalism .
in addition , in our case @xmath430 .
thus , at a given temperature the phonon population is much higher than that of the spin boson .
this also implies higher relaxation rate for the phonons .
if the above is true then the momentum relaxation of spin bosons can be viewed as a two - stage process with a bottle - neck .
first , the momentum waits the longest time for a spin - boson@xmath431phonon collision to get transfered to the phonon subsystem .
second , the phonons quickly dissipate the transferred momentum .
the first stage is the bottle - neck .
therefore , it controls the rate of the spin boson momentum dissipation .
this is why we do not need to know microscopic details about the phonon - phonon collisions .
it is enough to know that these collisions occur more often than the spin - boson scattering on the phonons .
the factor which might lead to violation of our assumption is exponential weakness of the phonon umklapp relaxation at low enough temperatures .
the umklapp relaxation rate contains a characteristic exponent @xmath432 , @xmath433 is a constant of order of unity.@xcite therefore , our approximation breaks down at very low temperature . roughly speaking
, this temperature corresponds to the point where the smallness of the spin - lattice coupling is comparable to the smallness of @xmath432 . in such a situation
the bottle - neck scheme described above will cease to work and it will be necessary to include a detailed description of the phonon - phonon collisions as well . in practice , however , the impurity scattering will become relevant at much higher temperatures .
our theory shows a very weak impurity concentration dependence of the temperature below which impurities control the transport : @xmath60 .
this implies that even a very weak disorder will result in a substantial @xmath58 .
we have verified our assumptions using the phenomenological parameters from the spin - chain material sr@xmath0cuo@xmath1 and obtained that the two - stage process approximation is very well justified , see sec .
[ kappa ] .
we would like to note that the problem of the spin transport in spin chains coupled to 3d phonons has been studied within a different approach@xcite and gives a different temperature dependence for the spin thermal conductivity : @xmath234 .
the essential difference between our approaches is our treatment of the phonon - bath as a source of the momentum dissipation .
if this process is neglected the `` normal '' transfer of the momentum from spins to phonons does not lead to degrading of the thermal current .
then , it is only the `` direct '' spin - phonon umklapp processes which will be contributing to the transport relaxation rate with a characteristic exponential dependence on temperature .
however , this could only work if the phonon system would be very reluctant to dissipate the momentum , which can only happen at the temperatures of the order of the `` phonon peak '' ( 30k for sr@xmath0cuo@xmath1 ) , well below the temperature of the maximum in the spin part of @xmath298 ( @xmath434k ) . _
final remarks .
_ in this paper we have studied the problem of anomalous heat transport in quasi-1d spin - chain systems coupled to the 3d phonon environment in the presence of weak disorder . we have derived a microscopic model of 1d bosonic spin excitations interacting with phonons and impurities using bosonization approach .
we have considered the spin - phonon scattering within the boltzmann equation for the spin - boson distribution function .
this equation has been solved in the limit of weak spin - lattice coupling and fast spin - boson excitations .
assuming that the phonon relaxation time @xmath22 due to phonon - phonon scattering is much shorter than the spin - phonon relaxation time @xmath19 we have obtained the transport relaxation rate for the spin bosons @xmath435 for @xmath179 and @xmath436 for @xmath56 .
we have shown that within our model the spin - phonon scattering alone is insufficient to render the thermal conductivity finite because the spin - phonon relaxation mechanism becomes ineffective for the spin bosons at low energies .
on the other hand , the 1d impurity scattering provides a natural cut - off scale for the low - energy spin bosons and results in the momentum - independent scattering rate @xmath437 .
thus , the impurity scattering removes the divergence of the conductivity .
we have calculated the thermal conductivity as a function of temperature and have shown that the low temperature transport is dominated by the impurity scattering while the high temperature transport is limited by both the impurity scattering and the spin - phonon collisions .
our main results are in a very good quantitative agreement with the available experimental data .
in implementing our approach we have also obtained an insight into various microscopic details of the problem .
this has allowed us to formulate several predictions and suggest future experiments as well , see sec .
[ kappa ] .
one of the predictions is the saturation of the spin - boson thermal conductivity at high temperatures : a non - trivial result due to impurity and spin - phonon scatterings and 1d nature of spin - bosons .
we would like to thank n. andrei for useful conversations , e. orignac for communications , and c. yu for insightful comments which lead us to writing sec .
[ meanfp ] .
we are also grateful to a. sologubenko and c. hess for sending experimental data .
this work was supported by doe under grant de - fg02 - 04er46174 , and by the acs petroleum research fund .
a.v.r . is grateful to the dynasty foundation of dmitrii zimin for support of this research .
the integral over @xmath95 in @xmath149 , eq . ( [ s1 ] ) , can be evaluated explicitly .
let us denote the integrands of ( [ s1 ] ) other than delta - functions by @xmath438 .
then , the integration over @xmath95 can be performed giving the following answer : @xmath439 where @xmath440 is a step function and @xmath3 is the debye temperature .
the step - functions @xmath441 in ( [ pint ] ) impose restrictions on the possible scattering states of the spin boson with the given momentum @xmath32 .
these restrictions are consequences of the conservation laws . because of this step - functions the integration variable @xmath106 in eqs .
( [ s1 ] ) and ( [ s2 ] ) for the collision integrals @xmath442 is bound to an interval @xmath443 smaller than the brillouin zone .
this interval , in general , depends on @xmath32 . to find @xmath444 we must solve : @xmath445 for @xmath446 .
graphical solution of these inequalities for @xmath447 is presented in fig .
[ fig_app ] as a lightly shaded area on @xmath448 plane . in this
figure lines @xmath449 and @xmath450 are bisectors @xmath451 . although not shown in fig .
[ fig_app ] , @xmath452 portion of the graphical solution can be obtained by an inversion of fig .
[ fig_app ] with respect to the point @xmath453 .
equivalently , we can express the solution as the following choice of the integration interval @xmath454 for @xmath455 : @xmath456 \cup [ -k+\sqrt{4(c_\ell / v)^2 k^2 + ( \theta_{d\ell}/v)^2 } , -(1 - 2c_\ell / v)k ] & for $ k > 0$,\cr [ -(1 - 2c_\ell / v)k , -k-\sqrt{4(c_\ell / v)^2 k^2 + ( \theta_{d\ell}/v)^2 } ] \cup [ k+\theta_{d\ell}/v , k ] & for $ k < 0 $ , } \end{aligned}\ ] ] where we use @xmath457 $ ] to denote the interval from @xmath458 to @xmath459 . for @xmath460 : @xmath461 & for $ k > 0 $ , \cr [ -(1 - 2c_\ell / v)k , k ] & for $ k < 0$. } \end{aligned}\ ] ] for the contour @xmath462 in the integral for @xmath463 : @xmath464 \cup [ -(1 + 2c_\ell / v)k , -k - \theta_{d\ell}/v ] & for $ k
> 0$,\cr [ k - \sqrt{4(c_\ell / v)^2 k^2 + ( \theta_{d\ell}/v)^2 } , ( 1 + 2c_\ell / v)k ] \cup [ k , k - \theta_{d\ell}/v ] & for $ k < 0$.}\end{aligned}\ ] ] the @xmath447 part of this solution corresponds to the darker shaded area in fig .
[ fig_app ] . to the state @xmath106 .
the processes in which the original spin boson decays into another spin boson and emits the phonon are represented by the light gray area .
for them @xmath465 as dictated by the energy conservation .
the dark gray area corresponds to the processes of phonon absorption by the spin boson . for them
lines @xmath449 and @xmath450 are bisectors @xmath467.,width=226 ] keeping in mind the above results we rewrite the collision integral in the following form : @xmath468 where @xmath469 the contour @xmath470 is the union @xmath471 . for @xmath460 we must integrate between @xmath472 . for @xmath473 we must integrate between @xmath474 and between @xmath475 . in both cases
a window of the width @xmath476 centered around @xmath477 must be deleted from the integration interval . in fig .
[ fig_app ] one can identify this window with the unshaded area around line @xmath450 .
we start by making the following observation . in general , the intervals @xmath470 in ( [ sell ] ) differ for different polarizations @xmath478 : the size of the gap around @xmath477 depends on the value of @xmath101 , so do @xmath479 which specify the outer limits of the integration intervals .
however , we will show later that these variations in @xmath470 do not change the integrals ( [ sell ] ) significantly .
therefore , it is convenient to introduce @xmath480 and @xmath481 . using this notation we write : @xmath482 as it was mentioned the integrals over @xmath483 in the above expressions are small .
hence , it is permissible to integrate over @xmath484 in both @xmath485 and @xmath486 . in such a situation
the summation over polarization @xmath96 can be done easily since in both @xmath485 and @xmath486 only the product @xmath487 depends on polarization .
thus : @xmath488 where @xmath489 and @xmath490 .
therefore , the collision integral itself is equal to : @xmath491 the first term of this integral can be calculated for small @xmath492 : @xmath493 which gives the spin - boson transport relaxation time : @xmath494 @xmath495 where @xmath496 is defined in eq .
( [ gamma0 ] ) .
one can see from eq .
( [ tau ] ) that the dominant scattering process is the absorption of a phonon with the characteristic energy @xmath497 as this energy range provides the major contribution to the integral .
if we neglect the second term of ( [ sapp ] ) as well as the corrections to the collision integral coming from @xmath483 the function @xmath143 can be found right away .
for example , if @xmath498 : @xmath499 this is the expression for odd ( in @xmath32 ) part of the spin boson distribution function . using it
we estimate the omitted term of ( [ sapp ] ) and show that this term is small : @xmath500 \frac{e^{-|k'|/k_t } - 1}{e^{-|k|/k_t } -1}f_{k'}^1.\nonumber\end{aligned}\ ] ] of three terms in the square brackets only the second gives non - vanishing contribution to the integral .
two others do not contribute because they are even in @xmath106 and function @xmath501 is odd .
therefore : @xmath502 where @xmath503 .
since the integral in ( [ ax ] ) diverges at small @xmath504 we must introduce small cut - off parameter @xmath505 .
such a cut - off is associated with the impurity scattering .
it is equal to the momentum at which the impurity scattering rate becomes comparable with @xmath506 .
for now it is enough to assume that @xmath505 is not too small so we can neglect the logarithm in the above formula . the expression ( [ ax ] )
is to be compared with ( [ tau ] ) . for @xmath498 :
@xmath507 in other words , at low temperatures the second term of ( [ sapp ] ) is small as long as @xmath508 .
the same is true for larger temperature ( @xmath509 ) as well .
we also need to estimate the integrals over @xmath483 to demonstrate that they are small .
first , let us consider the integral over @xmath510 .
this interval is localized near @xmath477 : @xmath511 \cup [ -(1 - 2c_t / v)k , -(1 - 2c_l / v)k].\end{aligned}\ ] ] it is non - vanishing as long as @xmath512 .
it is necessary to perform the integral ( [ sell ] ) with @xmath513 over @xmath510 : @xmath514 the absolute value of the last expression can be bounded from above by the following : @xmath515 the latter expression can be estimated as @xmath516 as long as @xmath517 .
second , we estimate the integral ( [ sell ] ) with @xmath518 over @xmath519 : @xmath520 \cup [ -|k| - \theta_{dt}/v , -|k| - \theta_{dl}/v].\end{aligned}\ ] ] it is clear without extensive calculations that both terms of ( [ sell ] ) are proportional to @xmath521 .
thus , the integral over this interval does not contribute significantly to the collision integral of the spin bosons . | we study the problem of heat transport in one - dimensional ( 1d ) spin - chain systems weakly coupled to three - dimensional phonons and impurities .
we consider the limit of fast spin excitations and slow phonons , applicable to a number of compounds of the cuprates family , such as sr@xmath0cuo@xmath1 , where the superexchange @xmath2 is much larger than the debye energy , @xmath3 . in this case
the umklapp scattering among the spin excitations is strongly suppressed for all relevant temperatures .
we argue that the leading scattering mechanism for the spin excitations at not too low temperatures is the `` normal '' ( as opposed to the umklapp ) spin - phonon scattering in which the non - equilibrium momentum is transferred from the spin subsystem to phonons where it quickly relaxes through the `` phonon bath '' . because of the lower dimensionality of the spin excitations it is only the momentum along the chains which is conserved in such a scattering .
we find that this effect leads to a particular momentum- and temperature - dependence of the spin - phonon relaxation rate valid for the broad class of low - dimensional spin systems .
subsequently we demonstrate that the spin - phonon relaxation mechanism is insufficient for the low - energy , long - wavelength 1d spin - chain excitations , which make the thermal conductivity diverge .
we complete our consideration by taking into account the impurity scattering , which in 1d cuts off the quasi - ballistic spin excitations and renders the thermal conductivity finite .
altogether , these effects yield the following spin - boson thermal conductivity behavior : @xmath4 at low temperatures , @xmath5 at intermediate temperatures , and @xmath6 at higher temperatures @xmath7 .
the saturation at higher temperatures is of rather non - trivial origin and we provide a detailed discussion of it .
our results compare very well with the existing experimental data for sr@xmath0cuo@xmath1 .
using our microscopic insight into the problem we propose further experiments and predict an unusual impurity concentration dependence for a number of quantities . |
angle - dependent magnetoresistance oscillations ( amro ) in the interlayer resistance of layered metals provide a means to determine fermi surface properties of such systems.@xcite maps of two and three dimensional fermi surfaces have been determined for a wide variety of materials.@xcite amro can also be used to infer information about possible anisotropies in the momentum space of fermi surface properties such as the scattering rate.@xcite in particular , recent amro experiments by abdel - jawad _ et al . _
@xcite reveal that in the overdoped cuprate tl@xmath0ba@xmath0cuo@xmath1 ( tl2201 ) the scattering rate contains an anisotropic piece that varies with the same symmetry as the superconducting gap and grows linearly with temperature .
the measured anisotropy in the scattering rate may have relevance to the origin of high temperature superconductivity in cuprate superconductors , as the strength of the anisotropic piece of the scattering increases with the critical temperature , @xmath2.@xcite scattering rate anisotropy has been detected in several other cuprates using angle - resolved photoemission spectroscopy ( arpes),@xcite optical conductivity@xcite and in - plane transport measurements.@xcite theoretical support for such anisotropic scattering comes from ideas about hot and cold spots on the fermi surface , first raised in the context of optical conductivity,@xcite and appears to arise naturally in dynamical mean field calculations on the two dimensional hubbard model.@xcite an advantage of amro over other techniques which use oscillations to obtain information about the fermi surface , such as de haas - van alphen and shubnikov - de haas oscillations , is that they can be measured at elevated temperatures , provided @xmath3 , the combination of the cyclotron frequency and the transport lifetime is large enough . to date , expressions used to fit amro data have been derived assuming zero temperature , which should work reasonably well for temperatures @xmath4 , the fermi temperature .
recently there have been relatively high temperature amro experiments for which corrections to the zero temperature approximation may be needed . for instance , the experiments in refs . were performed for @xmath5 up to 55 k , corresponding to a maximum thermal energy of approximately 0.02 of the fermi energy @xmath6 . at this temperature scale ,
one expects the zero - temperature approximation to work reasonably well in fits to amro .
however , more recent experiments@xcite have extended amro to temperatures as high as 110 k ( @xmath7 ) where finite temperature effects should be more relevant we show below that finite temperature corrections to amro can be quantitatively important at these temperatures . in this paper
we generalize existing zero temperature semi - classical amro formulae for layered metals with anisotropic fermi surface properties to finite temperatures .
the finite temperature expression for interlayer conductivity with anisotropic fermi surface properties is our main result .
we use this expression to generate numerically the amro expected in overdoped thallium cuprate for an isotropic scattering rate .
fitting to the simulated amro with zero temperature expressions reveals no significant anisotropic contribution to the scattering rate .
our results indicate that the inferred scattering rates in ref .
are robust to finite temperature corrections , but that finite temperature corrections will be required in fits to amro at higher temperatures .
this paper is structured as follows : in sec .
[ sec : calculationofamro ] we present our analytic calculations of amro . in sec .
[ sec : numerics ] we perform numerical calculations to check whether the finite temperature effects can masquerade as linear in @xmath5 anisotropic scattering . in sec .
[ sec : conclusion ] we summarize our results and conclude .
in this section we briefly review the calculation of the interlayer conductivity in a layered metal allowing for non - zero temperature and treat the cases of isotropic fermi surface properties ( for which we can make considerable analytic progress ) and anisotropic fermi surface properties separately .
we assume a simple model of a layered metal in which the @xmath8-axis is the weakly conducting direction and take a dispersion relation of the form @xmath9 where @xmath10 is the electron wavevector , @xmath11 is the dispersion in the @xmath12-@xmath13 plane , @xmath14 is the ( possibly anisotropic ) interlayer hopping term ( assumed to be small compared to the fermi energy ) , and the parameter @xmath8 is the distance between the conducting layers .
we treat the electrons semiclassically and calculate amro by solving the time independent and spatially uniform boltzmann equation in the relaxation time approximation : @xmath15 where @xmath16 is the electron distribution function , in terms of momentum @xmath17 , @xmath18 is the fermi - dirac distribution , @xmath19 is the lorentz force for a magnetic field @xmath20 and a weak electric field perpendicular to the conducting layers , @xmath21 , @xmath22 is the velocity , and @xmath23 is the transport lifetime
. in general @xmath23 may depend on both the electron momentum and energy .
we assume that the magnetic field @xmath24 is applied at an angle @xmath25 with respect to the @xmath26-axis and an angle @xmath27 with respect to the @xmath28 axis , as shown schematically in fig .
[ fig : metaldiag ] . is measured relative to the @xmath26-axis and the azimuthal angle @xmath27 relative to the @xmath28-axis , which lies in the conducting plane . ] for weak electric fields we can write @xmath29 , where @xmath30 . neglecting small terms in the electric field and the hopping , @xmath31 , we obtain @xmath32 where @xmath33 is the angular position in momentum space and @xmath34 is the energy .
we relabel @xmath35 for convenience , and define@xcite @xmath36 with @xmath37 and @xmath38 .
the in - plane dispersion relation @xmath39 determines @xmath40 and @xmath41 ; accordingly , if @xmath11 is anisotropic , @xmath40 , @xmath41 and @xmath42 will also be anisotropic in momentum space .
equation ( [ eqn : generalboltz ] ) holds for both isotropic and anisotropic fermi surfaces .
the formal solution to eq .
( [ eqn : generalboltz ] ) is @xmath43 where @xmath44 , \label{eqn : green}\ ] ] is the probability that an electron will travel around the fermi surface from an angle @xmath45 to @xmath33 without being scattered.@xcite with this expression we may calculate the current density @xmath46 and hence the interlayer conductivity @xmath47 . in sec .
[ sec : isofermi ] and sec .
[ sec : anisofermi ] we present the resulting finite temperature interlayer conductivity for both isotropic and anisotropic cases , respectively .
we first consider finite temperature effects for an isotropic layered metal , with isotropic scattering rate @xmath48 ( we allow @xmath23 to depend on @xmath34 ) , isotropic hopping @xmath31 and an isotropic dispersion,@xcite @xmath49 where @xmath50 is the effective mass .
this gives @xmath51 .
in the zero temperature limit the expression for the interlayer conductivity is@xcite @xmath52 , \label{eqn : cond}\ ] ] where @xmath53 ( @xmath23 , if energy dependent , is evaluated at the fermi energy ) and @xmath54 . to generalize this expression to finite temperature
we must integrate over energy in calculating @xmath47 .
the integral to be evaluated is @xmath55 , \nonumber \\ \label{eqn : isoenergy}\end{aligned}\ ] ] where the energy dependence of the scattering rate @xmath56 must be taken into account , @xmath57 , and @xmath58 . using a sommerfeld expansion@xcite and keeping terms to order @xmath59
, we arrive at the result @xmath60\right .
\nonumber \\ & & \hspace*{0.5cm}+~\frac{\pi^2}{6}(k_bt)^2\left.\left[\psi_0(\varepsilon_f ) + 2\sum_{s=1}^{\infty}\psi_s(\varepsilon_f)\right]\right\ } , \nonumber \\ \label{eqn : result}\end{aligned}\ ] ] where the chemical potential is @xmath61 and we introduce @xmath62 .
\label{eqn : psis}\ ] ] in fig .
[ fig : resvstheta ] we plot the interlayer resistivity @xmath63 ( normalized by @xmath64 ) , determined from eqs .
( [ eqn : cond ] ) and ( [ eqn : result ] ) for an energy - independent @xmath23 at several different temperatures .
we choose @xmath65 and @xmath66 ( where @xmath67 ) to match values used in plots in ref . and plot temperatures of 0 , @xmath68 , @xmath69 and @xmath70 .
deviations from the zero temperature result are mainly noticeable in the first few amro peaks , for @xmath71 .
interestingly , the minima seem much less affected by finite temperature than the maxima . to see why the maxima are affected more than minima we perform an asymptotic expansion of eq .
( [ eqn : result ] ) in the limit @xmath72 for an energy independent @xmath23 .
the derivation follows that for the @xmath73 case in ref . .
we find the correction to the zero temperature result to be ( when @xmath74 ) @xmath75 .
\label{eqn : asymcor}\end{aligned}\ ] ] the extrema of the resistivity occur at the yamaji angles,@xcite which satisfy @xmath76 where @xmath77 corresponds to a minimum ( maximum ) of the resisitivity @xmath63 .
[ note that there will be @xmath78 corrections to the yamaji angles , although as can be seen in fig .
[ fig : resvstheta ] , these are not large , and they are not important for our discussion here . ]
the convention used here for @xmath79 is that the @xmath80 minimum follows the @xmath80 maximum and @xmath79 begins at zero .
we can calculate the @xmath81 correction to the extrema of the resistivity using eq .
( [ eqn : asymcor ] ) .
we find that at maxima the relative change in the resistance is @xmath82 , \end{aligned}\ ] ] and at minima the relative change in resistance is @xmath83 .\end{aligned}\ ] ] versus @xmath25 for an isotropic fermi surface with @xmath84 and @xmath85 . ] in general the numerators in the two expressions will be comparable , but the denominator in the minima expression is always greater than 2 , whereas the denominator in the maxima expression can become very small when @xmath86 becomes large , leading to a much stronger reduction in the resistance at the maxima than increase at the resistance minima .
[ it should be noted that assuming temperature independent scattering , as we do in this example , is somewhat artificial ( in general one expects the scattering rate to have a @xmath87 contribution in a fermi liquid ) , but it serves to illustrate the differing effects of temperature on resistance minima and maxima . ]
we now turn to consider finite temperature effects for an anisotropic fermi surface .
we allow @xmath88 , @xmath23 , @xmath40 , @xmath41 and @xmath31 to vary with @xmath33 and @xmath34 .
the zero temperature interlayer conductivity for a layered metal with anisotropic fermi surface properties was calculated by kennett and mckenzie @xcite to be @xmath89 where @xmath90 , @xmath91 is the probability that an electron will make a full cyclotron orbit about the fermi surface before being scattered and @xmath92 where we have defined @xmath93 as the direction of the component of the magnetic field in the @xmath28-@xmath94 plane , and @xmath95 , where @xmath96 is the magnitude of the fermi wavevector .
generalizing eq .
( [ eqn : result1 ] ) for the interlayer conductivity to finite temperatures is similar to the isotropic case , and again we integrate over energy : @xmath97 the factor @xmath98 is as in eq .
( [ eqn : green ] ) , and the energy - dependent @xmath99 is @xmath100 where @xmath101 is the magnitude of the electron wavevector .
allowed values of @xmath102 are determined from the dispersion relation . equation ( [ eqn : result2 ] ) is our main result and is independent of whether the interlayer hopping is coherent ( assumed here ) or weakly incoherent.@xcite the only difference in amro for the two models of transport occurs for angles near @xmath103.@xcite in order to make analytic progress with eq .
( [ eqn : result2 ] ) we need to specify the energy and @xmath33 dependence of various fermi surface properties . in general ,
amro must be calculated numerically , although under certain conditions we may derive asymptotic expressions .
to do so , we note that if @xmath88 never approaches zero , the largest contribution to the integrand of eq .
( [ eqn : result2 ] ) will be from near @xmath104 , where the derivative of the fermi - dirac function is sharply peaked .
we can then perform a sommerfeld expansion about this point , as in the isotropic finite temperature case .
the zeroth order term gives the zero temperature result evaluated at @xmath105 instead of @xmath106 . under the conditions @xmath107 we can evaluate the zeroth order term using a stationary phase calculation.@xcite to second order in @xmath108 we can then write the temperature dependent conductivity , valid near @xmath103 , for a fermi surface with @xmath109 symmetry as @xmath110\nonumber \\ & & + ~\frac{\pi^2}{6}ebs_0\cos\theta(k_bt)^2\omega(\varepsilon_f ) , \label{eqn : anisotemp}\end{aligned}\ ] ] where @xmath111 satisfies @xmath112 = 0 $ ] , @xmath113 $ ] , and @xmath114 \nonumber \\ & \simeq & \frac{ebs_0\cos\theta}{c\tan\theta}\frac{d^2}{d\varepsilon^2}\left\{\frac{2\pi|\eta(\varepsilon,\phi_0)|^{-1}}{1-p(\varepsilon ) } \left(\frac{t_c(\varepsilon,\phi_0)}{\omega_0(\varepsilon,\phi_0)}\right)^2 \left[1+p(\varepsilon)+2\sqrt{p(\varepsilon)}\sin\left(2c\tan\theta \kappa(\varepsilon,\phi_0)\cos(\varphi-\phi_0)\right)\right]\right\}. \nonumber \\ \label{eqn : correction}\end{aligned}\ ] ]
we now outline the numerical scheme we use to test whether the temperature dependence of the conductivity affects the anisotropy obtained from fits to the scattering rate @xmath48 using zero temperature expressions for the conductivity .
our approach is as follows : we use the tight - binding dispersion relation inferred from arpes@xcite to simulate the amro data we expect to observe at several different temperatures and azimuthal field angles @xmath27 , assuming an isotropic scattering rate of the form predicted by fermi liquid theory .
we then fit to this data using the same procedure used to fit experimental data in ref . .
first , we allow all parameters to vary in fitting the lowest temperature data ( using the zero temperature expression for the interlayer resistivity ) .
second , the higher temperature data is fitted assuming only the scattering rate is temperature dependent .
if the scattering rate anisotropy is a fitting artifact , fits to amro data simulated with finite temperature expressions for the conductivity should yield results similar to the fits to the real data : an anisotropic contribution to the scattering rate that varies as @xmath115 . the tight binding dispersion relation determined by plat _
et al_.@xcite is : @xmath116 where the wavenumbers are measured in units of the in - plane lattice constant @xmath117 , the @xmath118 are hopping parameters and @xmath105 is the chemical potential .
the values of the parameters are @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath123 and @xmath124 , all in ev .
we use @xmath11 to calculate @xmath125 and @xmath102 as functions of @xmath33 numerically , for several energies .
we fit the output data at several energies to give a functional form of @xmath125 and @xmath102 as functions of @xmath33 and @xmath34 .
the fermi surface for the dispersion eq . ( [ eqn : tightbinding ] ) exhibits a @xmath109 symmetry , and so @xmath126 and @xmath127 also have this symmetry .
this allows us to fit the data to a truncated fourier series with only @xmath128 , @xmath129 , terms present.@xcite we find that @xmath127 may be accurately represented by the form @xmath130 and @xmath126 may be parametrized as @xmath131 for the energy range we are interested in the energy dependence of each of the coefficients can be well approximated as linear , @xmath132 .
contributions to the integral [ eq . ( [ eqn : result2 ] ) ] from energies far from the fermi energy are negligible .
the fitting parameters for @xmath102 and @xmath125 are given in table [ tb : coeffs ] .
.fitting parameters for @xmath126 and @xmath127 .
[ tb : coeffs ] [ cols=">,>,>,>,>,>,>,>",options="header " , ] the remaining functional inputs we need to calculate the amro are the interlayer hopping @xmath133 and the scattering rate @xmath134 .
we assume the standard isotropic scattering rate for @xmath135:@xcite @xmath136 .
\label{eqn : flscatter}\ ] ] we estimate the parameters @xmath137 and @xmath138 for @xmath48 from fits to the isotropic part of the scattering rate in ref . , giving @xmath139 ; @xmath137 always appears as part of a product @xmath140 , where @xmath141 is the constant contribution to @xmath127 at the fermi energy , and we find that @xmath142 .
we assume the hopping is independent of energy and use the same expression for the hopping term as in ref . :
\label{eqn : hopping}\ ] ] the parameters @xmath144 and @xmath145 are related to each other by the symmetry of the crystal lattice , which requires @xmath146 and hence @xmath147 . due to imperfections in the crystal
, this relation may not hold exactly , and we allow both parameters to vary during fitting .
the input values are @xmath148 and @xmath149 .
the constant @xmath150 enters the overall normalization and does not need to be specified in our calculation . with explicit functional forms for all functions appearing in eq .
( [ eqn : result2 ] ) , we may perform the angular and energy integrals . due to the fact that @xmath151 decays quickly away from @xmath104 we only integrate over the energy from @xmath152 to @xmath153 to cover the range of energy that contributes to the integral , and we checked that our results were independent of the energy range for this choice of integration interval . the values of @xmath27 we use are 0 , 20 , 28 , 36 and 44 degrees and we generate amro for @xmath154 , where @xmath155 is @xmath156 .
these correspond to temperatures from 3 k to 113 k. the output is normalized such that @xmath157 .
the simulated amro data are shown in fig .
[ fig : fakeamro ] . used
are 0 , 20 , 28 , 36 and 44 degrees .
note the change in vertical axis limits in the last plot .
the plots shown correspond to temperatures of 2.8 k , 5.7 k , 11.3 k , 28 k , 56 k and 113 k. ] we fit to the simulated amro data using eq .
( [ eqn : hopping ] ) for @xmath133 and parameterize the scattering rate as@xcite @xmath158 .
\label{eqn : scatterfit}\ ] ] the fermi wavevector and cyclotron frequency are parametrized as @xmath159 , \label{eqn : wavefit}\ ] ] and @xmath160 .
\label{eqn : omegafit}\ ] ] we use the 2.8 k data to fit to the parameters @xmath161 , @xmath162 , @xmath163 , @xmath164 , @xmath165 , @xmath144 and @xmath145 . for fits at temperatures above 2.8 k
we assume that only the scattering rate is temperature dependent , and fit only to the parameters @xmath164 and @xmath161 .
the scattering rate parameters @xmath166 and @xmath167 are plotted as a function of temperature in fig .
[ fig : simdatafits ] . a quadratic fit @xmath168 to @xmath166 yields @xmath169 and @xmath170 k , agreeing with the parameter values used to simulate the amro .
the fit values obtained are @xmath171 , @xmath172 , @xmath173 , @xmath174 and @xmath175 , which are within @xmath176 of the input parameter values used in the simulation .
we find that not only do the fits to the simulated data fail to reproduce the @xmath115 term observed in the experiments , we observe essentially no temperature dependence of the anisotropy parameter @xmath161 , suggesting that if the scattering rate were in fact isotropic the fitting procedure used by abdel - jawad _ _ et al.__@xcite would not have yielded anisotropic scattering .
hence , the anisotropic contribution to the electron - electron scattering rate measured in recent amro experiments is not an artifact of the fitting procedure , and the system does deviate from standard fermi liquid behavior . and @xmath167 as functions of temperature . ]
in this work we present expressions for the interlayer conductivity of layered metals with either an isotropic or anisotropic fermi surface , allowing for thermal effects . we used the expression for the conductivity for a layered metal with anisotropic fermi surface properties to simulate the amro expected an isotropic scattering rate consistent with fermi liquid theory .
fitting to the simulated data using the same procedure used to fit the experimental data in ref .
does not reproduce the anisotropic @xmath115 piece of the scattering rate .
hence , we have confirmed that including finite temperature effects whilst taking into account anisotropy in the dispersion does not affect the conclusion there there is an anisotropic contribution to the scattering rate of thallium cuprate .
the theory presented here is applicable to any layered metal in which amro may be observed , and provides the means to analyze amro data for temperatures at which @xmath108 grows large enough that finite temperature effects on amro become significant .
this may allow for a more precise interpretation of current amro data at higher temperatures , for example , in cuprate experiments@xcite where the maximum value of @xmath108 is @xmath177 or in na@xmath178coo@xmath0 , which has a fermi energy around 250 mev,@xcite and amro experiments have been performed at temperatures as high as @xmath179.@xcite this theory will also enable accurate quantitative analysis of future high temperature amro experiments .
we thank r. mckenzie and n. hussey for helpful discussions on amro . this work was supported by nserc .
m. p. kennett and r. h. mckenzie , phys .
b , * 76 * , 054515 ( 2007 ) .
m. abdel - jawad , j. g. analytis , l. balicas , j. p. h. charmant , m. m. j. french and n. e. hussey , phys .
lett . * 99 * , 107002 ( 2007 ) .
m. plat , j. d. f. mottershead , i. s. elfimov , d. c. peets , r. liang , d. a. bonn , w. n. hardy , s. chiuzbaian , m. falub , m. shi , l. patthey , and a. damascelli , phys .
* 95 * , 077001 ( 2005 ) ; d. c. peets , j. p. f. mottershead , b. wu , i. s. elfimov , r. liang , w. n. hardy , d. a. bonn , m. raudsepp , n. j. c. ingle and a. damascelli , new . j. phys * 9 * , 28 ( 2007 ) .
a. kaminski , h. m. fretwell , m. r. norman , m. randeria , s. rosenkranz , u. chatterjee , j. c. campuzano , j. mesot , t. sato , t. takahashi , t. terashima , m. takano , k. kadowaki , z. z. li , and h. raffy , phys .
b * 71 * , 014517 ( 2005 ) .
j. chang , m. shi , s. pailhs , m. mansson , t. claesson , o. tjernberg , a. bendounan , y. sassa , l. patthey , n. momono , m. oda , m. ido , s. guerrero , c. mudry , and j. mesot , phys .
b * 78 * , 205103 ( 2008 ) . for the purposes of illustration we ignore possible energy dependence of the interlayer hopping @xmath31 for an isotropic fermi surface , but we do include this possibility in our more general analysis of a layered metal with anisotropic fermi surface properties . | studies of interlayer transport in layered metals have generally made use of zero temperature conductivity expressions to analyze angle - dependent magnetoresistance oscillations ( amro ) .
however , recent high temperature amro experiments have been performed in a regime where the inclusion of finite temperature effects may be required for a quantitative description of the resistivity .
we calculate the interlayer conductivity in a layered metal with anisotropic fermi surface properties allowing for finite temperature effects .
we find that resistance maxima are modified by thermal effects much more strongly than resistance minima .
we also use our expressions to calculate the interlayer resistivity appropriate to recent amro experiments in an overdoped cuprate which led to the conclusion that there is an anisotropic , linear in temperature contribution to the scattering rate and find that this conclusion is robust . |
general - purpose quantum computers hold the promise of achieving quantum speed - ups in many problems of practical importance , unmatched by any known classical methods @xcite . while the prospect of such speed - ups is exciting , a growing realization is the extreme difficulty of achieving the levels of precision and control required for building truly scalable , fault - tolerant quantum hardware . as an intermediate step towards this goal ,
several recent proposals have suggested the development of special - purpose quantum devices which achieve so - called quantum supremacy " in certain tasks @xcite . instead of solving general computational problems , these devices instead sample from probability distributions widely believed to be impossible to simulate efficiently using classical means .
the recent explosion of proposals for such classically intractable sampling devices has begun to be matched by actual demonstrations of sampling in the laboratory @xcite , although so far still at small enough scales to allow for exact classical simulation .
an important question regarding such proposals is how far , and in what manner , we can reduce the resources required to exhibit and certify a genuine quantum advantage in sampling .
the boson sampling protocol @xcite shows that such quantum advantage can be achieved using simple linear optical devices and single - photon detectors . however , there are many challenges facing a realistic implementation of boson sampling , including the parallel generation of many single photons , the precise timing constraints on these photons , and the robust and accurate arrangement of the required beam splitters and phase shifters .
an alternative proposal which circumvents this bottleneck is the family of instantaneous quantum polynomial - time ( iqp ) protocols @xcite , where sampling distributions arise from single - qubit measurements on the output of low - depth commuting quantum circuits . if a quantum device can prepare sampling distributions associated with any unitary within a circuit family , then that process would be classically intractable under reasonable conjectures from complexity theory .
furthermore , the commuting nature of these quantum circuits means that they can potentially be engineered to run in constant time , maximally avoiding the threat of environmental noise and decoherence .
however , a practical issue which arises here is the extreme difficulty of engineering the arbitrary long - range interactions needed for such a constant time implementation .
while these long - range interactions can be simulated by bringing distant qubits together using @xmath0 gates before applying local entangling operations , this process would introduce a new bottleneck , the growing time required to shuttle qubits between local interaction regions . in the absence of quantum error correction , the growing influence of decoherence
would quickly degrade the quality of our sampling distributions , making this straightforward implementation likely untenable for practical demonstrations of quantum supremacy . in this paper ,
we show how nonadaptive measurement - based quantum computation ( mqc ) @xcite can be used to sample from the distributions associated with iqp circuits , while at the same time verifying the classical intractability of this sampling process .
our protocol uses a fixed resource state preparable by a constant - depth local circuit , which is then nonadaptively measured at each site in the pauli @xmath1 , @xmath2 , or @xmath3 bases .
the setting of nonadaptive mqc allows us to replace the time complexity present in local iqp circuits ( with @xmath0 gates ) by a spatial overhead in our resource state , which results in a protocol with constant runtime and local interactions .
the cost of this nonadaptivity is a fundamental randomness in the distributions prepared by our protocol , arising from random mqc byproduct operators .
this leads each sample in our protocol to be obtained with high probability from a different sampling distribution every time .
surprisingly , we show that this inherent randomness has no impact on the hardness of our protocol , which remains classically intractable under the same assumptions as in @xcite . what s more , we show that these random byproduct operators actually simplify our implementation relative to a direct circuit - based counterpart , revealing an inherent advantage of mqc for quantum sampling protocols .
we further show that by simply changing the single - qubit pauli measurements used to obtain sampling statistics , we can rigorously verify the classical intractability of our sampling .
our verification scheme is inspired by the ground state certification protocol of @xcite , but uses the special form of our iqp sampling distributions to replace the nonlocal operations required for general hamiltonian measurements with measurements of single - qubit pauli operators .
this lets us switch between sampling and measurement by a simple change in single - qubit measurement bases , allowing our procedure to achieve a robust demonstration of quantum supremacy capable of efficiently detecting any errors which could potentially harm the correctness of our sampling distributions .
our protocol is closely related to that of @xcite , as it constitutes a faithful translation of their circuit - based iqp sampling into the context of mqc .
however , we show that this translation itself contains several surprises , ultimately revolving around the nontrivial interface of mqc byproduct operators with classically intractable sampling . at first glance , our protocol has much in common with @xcite , which also use nonadaptive mqc to perform classically intractable sampling and verification . upon further investigation however ,
the different protocols are seen to utilize completely different mechanisms for demonstrating quantum supremacy . while using a more involved resource state than the ising - like states of @xcite , the design of our protocol allows for a convenient duality between sampling and verification , in which sampling and verification are both implemented using only single - qubit measurements on our output sampling state . in section [ sec : background ] , we review the relevant theory behind iqp sampling , verification , and mqc . in section [ sec : protocol ]
we present our protocol for preparing , sampling from , and verifying different classically intractable sampling distributions using pauli measurements on a model resource state @xmath4 . in section [ sec : outlook ]
we comment on the features unique to our protocol , and outline future directions for our work . a brief comparison of our proposal to other proposals within the rapidly growing field of classically intractable sampling can be found in appendix [ sec : comparison ] , with detailed proofs of the classical intractability and verification of our sampling protocol found in appendices [ sec : preparation ] , [ sec : sampling ] and [ sec : verification ] .
in the iqp sampling protocols of @xcite , a sampling state @xmath5 is first prepared using an @xmath6-qubit diagonal unitary circuit @xmath7 , and is then measured everywhere in the pauli @xmath1 basis to obtain a random outcome @xmath8 . in the above , @xmath9 denotes the @xmath10 eigenstate of @xmath1 , @xmath11 the single - qubit hadamard operator , @xmath12 a bit string of length @xmath6 , and @xmath13 the corresponding @xmath3 basis product state .
if @xmath7 is chosen from an appropriate family of diagonal unitaries , then @xcite shows that the act of sampling from the distribution @xmath14 is impossible to perform in polynomial time using a classical computer , assuming the widely conjectured non - collapse of the polynomial hierarchy of complexity theory @xcite .
more generally , we use the phrase classically intractable sampling to mean any sampling protocol which shares this property of being impossible to simulate classically ( given the non - collapse of the polynomial hierarchy ) , possibly in the presence of some allowable error and under the assumed truth of additional mathematical conjectures .
we now choose the @xmath6-qubit unitary gates @xmath7 above to be parameterized by @xmath6-bit binary functions @xmath15 , where @xmath16 denotes the finite field of binary numbers .
the functions @xmath17 set the eigenvalues of @xmath7 as @xmath18 where @xmath19 .
when applied to @xmath20 , this results in the sampling state @xmath21 we can alternately describe @xmath22 as the unique state satisfying the @xmath6 ( nonlocal ) stabilizer relations @xmath23 for @xmath24 , where @xmath25 and the polynomial @xmath26 is equal to the difference @xmath27 because addition in @xmath28 is modulo 2 , it is easy to verify that @xmath29 is always independent of the value of @xmath30 .
we now restrict our binary functions to be cubic polynomials , so that @xmath31 can be written in the form @xmath32 for some binary coefficients @xmath33 , @xmath34 , and @xmath35 .
these are generated by linear , quadratic , and cubic monomials , whose associated diagonal unitary gates are @xmath36 , @xmath37 ( controlled-@xmath3 ) , and @xmath38 ( controlled - controlled-@xmath3 ) . in the following , any references to polynomials will be understood to refer specifically to binary polynomials .
we will use @xmath39 , @xmath40 , and @xmath41 to denote homogeneous polynomials , for which the only nonzero coefficients are of the form @xmath33 , @xmath34 , or @xmath35 , respectively .
similarly , @xmath42 and @xmath43 will denote polynomials for which all @xmath44 or all @xmath45 , respectively
. it will be convenient in the following to interpret @xmath6-bit vectors @xmath46 as linear polynomials of @xmath6 variables , which act as @xmath47 this is useful in giving the probability of different sampling outcomes , as the probability of obtaining any given @xmath48 when @xmath22 is measured in the @xmath1 product basis is @xmath49 @xmath50 refers here to the square of @xmath51 , the ( signed ) difference between the fraction of inputs yielding @xmath52 and @xmath53 .
@xmath51 is known to be # p - hard to compute for arbitrary cubic polynomials @xmath17 @xcite , and we will see that this hardness underlies the classical intractability of our sampling protocol . it is shown in @xcite that estimating the quantity @xmath50 up to @xmath54 multiplicative error , so that @xmath55 for arbitrary cubic polynomials @xmath17 , is # p - hard , mirroring the difficulty of computing @xmath51 .
this hardness leads to a similar finding as in @xcite , that exactly sampling from the cubic polynomial distributions @xmath56 defined in eq .
( [ eq : distribution ] ) is classically intractable . in particular , assuming the existence of a classical randomized algorithm which can efficiently sample from any of the distributions @xmath56 lets a technique called stockmeyer approximate counting @xcite be used to estimate the probabilities @xmath57 up to @xmath54 multiplicative error , and thus to solve arbitrary # p problems .
while stockmeyer counting is an unphysical process which can not be implemented with realistic classical or quantum computers , it can be carried out at a finite level of the polynomial hierarchy , and the hardness of # p problems for this hierarchy then leads to its collapse .
details of this process can be found in appendix [ sec : sampling ] . on the other hand ,
we have seen that these distributions appear naturally as the output distributions of the iqp sampling protocol described above , which allows us to interpret a concrete implementation of this protocol as a provable demonstration of quantum supremacy " . while straightforward and conceptually compelling , a major limitation of the above result is the impossibility of verifying that any realistic quantum protocol is sampling from _ exactly _ the ideal distribution @xmath56 @xcite . in order to demonstrate quantum supremacy in a more realistic setting , an alternate proof is given in @xcite which shows the classical intractability of sampling from any distribution @xmath58 which is variationally close to @xmath59 .
variationally close means here that the statistical distance between @xmath58 and @xmath59 is bounded by a constant @xmath60 , so that @xmath61 in @xcite a value of @xmath62 was shown to be sufficient for classically intractable sampling , which in appendix [ sec : sampling ] we show can be relaxed to @xmath63 ( although both values rely on the particular value of @xmath64 appearing in conjecture [ conj ] below ) .
this result is appealing from a practical standpoint , as the quantity @xmath65 can be efficiently estimated in experiments involving quantum sampling distributions . on the other hand , the above average - case " sampling result relies upon one additional complexity theoretic conjecture : [ conj ] let @xmath17 be an arbitrary cubic polynomial of the form given in eq .
( [ eq : cubic_poly ] ) .
then it is # p - hard to efficiently calculate an estimate @xmath66 of @xmath50 for which @xmath67 , on at least @xmath68 of polynomials @xmath17 .
intuitively , this conjecture states that even when our estimates @xmath66 are allowed to fail with some finite probability @xmath64 , corresponding to realistic errors in our sampling distributions @xmath58 , the problem of estimating @xmath50 on the remaining instances is still # p - hard .
while this reliance on an additional unproven conjecture is nt desirable , an analogous conjecture is required for every known average - case classically intractable sampling result , and thus is nt any special demerit of @xcite .
the techniques of @xcite can be used to efficiently verify the condition @xmath69 when @xmath58 arises from measurements on experimentally prepared quantum sampling states @xmath70 , which approximate our intended @xmath22 .
given @xmath70 , we can perform measurements of the nonlocal hermitian stabilizers @xmath71 defined in eq .
( [ eq : hami ] ) , which will always yield the outcome @xmath10 in the ideal case where @xmath72 . in more general cases ,
a sufficiently accurate empirical estimate of these @xmath6 observables @xmath71 can be converted into a bound on the statistical distance between the distributions @xmath58 and @xmath59 .
if the average @xmath73 is sufficiently close to @xmath10 so as to guarantee @xmath69 , then we can confidently conclude that our quantum protocol is performing classically intractable sampling .
we will soon show that the nonlocal measurements of @xmath71 can actually be replaced with single - qubit @xmath1 and @xmath3 measurements , which allows this verification to be done entirely in the setting of mqc .
mqc is a means of carrying out computation using only single - qubit measurements on a fixed many - body resource state . in this framework
, the choice of measurements made on local regions of our resource state determines logical operations which are applied to encoded logical qubits , while simultaneously teleporting these qubits to adjacent unmeasured sites .
the randomness of quantum measurement leads the outcomes of these measurements to determine a so - called byproduct operator , which acts as a random correction to the overall logical operation .
for example , in figure [ fig : gadgets]a we show the standard protocol for teleporting one logical qubit within the mqc quantum wire known as the 1d cluster state . given two successive @xmath1 measurements with outcomes @xmath74 and @xmath75 , the resultant logical operation is @xmath76 , showing the intended logical unitary to be the identity and the byproduct operator to be a random pauli @xmath77 . in figure [ fig : gadgets]b we show a gadget for performing the two - qubit @xmath0 operation on logical qubits , for which the byproduct operator is a random two - qubit pauli operator . in both of these examples ,
the collection of operators appearing as byproducts for arbitrary measurement outcomes form a closed group ( up to global phase ) of finite size , referred to as a byproduct group .
an mqc protocol is said to be adaptive if the choice of measurement in some region of our resource state depends on the outcome of measurements made in another region .
adaptation can be seen as a means of ensuring that the byproduct group associated with a large computation remains finite ( for example , contained within the @xmath6-qubit pauli group ) , whereas the use of nonadaptive mqc with arbitrary single - qubit measurements will generally lead to a byproduct group of unbounded size . on the other hand ,
nonadaptive mqc computations can always be implemented in constant time by performing all measurements simultaneously , a serious advantage in the absence of quantum error correction . within the usual scheme for universal mqc using resource states built from @xmath78 gates ,
nonadaptive single - qubit pauli measurements are associated with byproduct groups formed from pauli operators , and implement logical operations contained within the clifford group .
the clifford group is defined as those unitaries @xmath79 which preserve the pauli group under conjugation , so that @xmath80 is a product of pauli operators whenever @xmath81 is .
the evolution of pauli eigenstates under the clifford group is known to be efficiently simulable using classical means @xcite , which means that non - clifford operations are necessary for demonstrating quantum supremacy . per measurement (
not shown ) .
mathematically , this leads our measurement outcomes @xmath82 to occur uniformly randomly .
( a ) 1d cluster state wire of length 2 , where solid lines indicate @xmath78 formation unitaries .
measuring @xmath1 on two sites implements the identity , with a uniformly random pauli byproduct group .
( b ) planar mqc gadget for implementing nonplanar wire crossings .
measuring @xmath1 on 6 sites implements @xmath0 , with a byproduct group of uniformly random two - qubit pauli operators .
( c ) non - clifford gadget for conditional @xmath83 , where blue triangles indicate @xmath83 gates used to form the gadget .
measuring @xmath2 on 3 non - logical control sites ( red ) gives @xmath83 on sites @xmath84 ,
@xmath85 , and @xmath86 , whereas measuring @xmath3 on these sites instead gives the identity . in both cases , the teleportation is trivial ( output and input sites coincide ) , while the byproduct group is a product of uniformly random @xmath78 s between @xmath84 and @xmath86 , and between @xmath85 and @xmath86.,scaledwidth=49.0% ] in figure [ fig : gadgets]c , we give an example of an mqc gadget which implements a non - clifford @xmath83 gate when nonadaptive pauli measurements are applied .
this gadget , which will be utilized in our classically intractable sampling protocol below , is itself formed from non - clifford @xmath83 gates , and has a byproduct group containing non - pauli @xmath78 gates .
a similar gadget was shown in @xcite to enable universal mqc using only pauli measurements , but with adaptation of measurement bases so as to avoid a byproduct group of unbounded size . in our mqc sampling protocol
below , we will show that restricting our logical operations to those generating sub - universal quantum computation will allow us to avoid this use of adapation , while still maintaining a byproduct group of finite size .
in fact , we will find that this non - pauli byproduct group actually leads to a simplification in our protocol relative to circuit - based counterparts .
our mqc implementation of the classically intractable sampling protocol of @xcite uses nonadaptive pauli measurements to prepare , sample from , and verify the @xmath6-qubit sampling states @xmath22 described above , for arbitrary cubic polynomials @xmath17 .
our protocol uses a 2d resource state @xmath4 which is capable of preparing any sampling state @xmath22 using only single - qubit pauli measurements .
@xmath4 is constructed from the teleportation , @xmath0 , and @xmath83 gadgets described in section [ sec : mqc ] , which are configured to implement any of the iqp circuits @xmath87 associated with arbitrary homogeneous cubic polynomials @xmath39 .
the choice of @xmath39 is determined by the choice of pauli measurement basis applied to each @xmath83 gadget in @xmath4 . by virtue of the byproducts arising from our nonadaptive mqc implementation ,
our output sampling states end up being random @xmath22 where @xmath88 is a sum of the intended @xmath39 , along with random quadratic and linear polynomials @xmath40 and @xmath41 . owing to this randomness in @xmath42 ,
we are unable to deterministically prepare any fixed sampling state @xmath22 . despite this fundamental indeterminism
, we will show how the act of sampling from randomly prepared @xmath22 with @xmath1 measurements at the final stage of our protocol remains classically intractable , even in the presence of realistic noise which leads our output sampling distributions to be some imperfect @xmath58 .
we state the classical intractability of our protocol , and the precise conditions which guarantee this , as theorem [ thm : one ] .
[ thm : one ] assume the validity of conjecture [ conj ] and the non - collapse of the polynomial hierarchy .
if the distributions @xmath89 arising from our mqc sampling protocol are close on average to the distributions @xmath57 defined in eq .
( [ eq : distribution ] ) , meaning that the average @xmath90 norm over all @xmath17 meets the experimental threshold @xmath91 , then our protocol is impossible to efficiently simulate using a classical computer , i.e. is classically intractable .
our protocol for classically intractable sampling is divided into two stages : preparation of the random sampling state @xmath22 and sampling / verification measurements on @xmath22 ( see figure [ fig : protocol ] ) . in the preparation stage , we use @xmath92 single - qubit measurements of pauli @xmath1 , @xmath2 , and @xmath3 on @xmath4 with outcomes @xmath93 to prepare the @xmath6-qubit state @xmath94 associated with a @xmath95-dependent polynomial @xmath96 .
these measurements are chosen to implement the unitary @xmath87 by means of a depth @xmath97 quantum circuit built from local @xmath83 and @xmath0 gates .
the @xmath83 gates in this ideal circuit are applied conditionally as @xmath98 , depending on the coefficients of @xmath39 , with teleportation and @xmath0 gates used before each application to move qubits @xmath99 , @xmath100 , and @xmath101 into the same region .
the application of these conditional @xmath83 s is structured within three nested levels of iteration , which together apply all @xmath102 three - body terms in the lexicographic order of the triples @xmath103 , where @xmath104 .
loop i , the lowest level of iteration , involves fixing qubits @xmath99 and @xmath100 in a designated interaction region , then successively cycling the remaining qubits @xmath105 through this region .
@xmath98 is applied in turn to each triple , until all triples @xmath103 with fixed @xmath99 and @xmath100 have been processed in this manner .
loop ii , the next level of iteration , involves successively replacing qubit @xmath100 by qubit @xmath106 , then repeating loop i for all qubits @xmath107 until all triples @xmath103 with fixed @xmath99 have been processed .
finally , loop iii involves successively replacing qubit @xmath99 by qubit @xmath108 , in the process shifting the location of the interaction region , and repeating loop ii for all qubits @xmath109 until @xmath98 has been applied to all triples of qubits .
the resulting unitary operation is clearly @xmath87 .
while the simple circuit described above is only capable of producing sampling states @xmath110 associated with homogeneous cubic @xmath39 , our mqc implementation utilizes random byproduct operators to implement the remaining quadratic and linear terms required for the preparation of arbitrary @xmath22 .
this reveals a simplification within nonadaptive mqc compared to a direct circuit - based counterpart , which would require additional @xmath78 and @xmath3 gates to implement @xmath7 for arbitrary @xmath17 .
each of the conditional operations @xmath98 is implemented using the @xmath83 gadget shown in figure [ fig : gadgets]c , which is measured in @xmath2 if @xmath111 and @xmath3 otherwise . for either choice of measurement
, the non - clifford nature of these gadgets leads the resultant byproduct operators to consist of non - pauli @xmath78 gates , which generate random quadratic terms in the output @xmath22 . because our logic gates and byproduct operators are made up of @xmath1 and the diagonal @xmath3 , @xmath78 , and @xmath83 gates , which together form a closed ( non - universal ) gate set under multiplication , the byproduct group associated with our computation
will always remain finite .
the @xmath83 gadgets used in our protocol are embedded in regular intervals in @xmath4 , and are then connected together using 1d cluster wires and @xmath0 gadgets , which simulate the movement of qubits utilized in our ideal quantum circuit described above .
these cluster wires and @xmath0 gadgets are always measured in @xmath1 , which leads to a product of random pauli @xmath1 and @xmath3 byproduct operators .
the @xmath3 byproducts eventually end up generating random linear terms in the output state @xmath22 , while the @xmath1 byproducts can be commuted backwards in our circuit , to eventually be annihilated on the initial @xmath20 which our logical quantum circuit is applied to .
this commutation of @xmath1 byproduct operators induces conditional @xmath112 and @xmath113 byproduct operators arising from prior @xmath83 gadgets , which results in additional randomness in the overall byproduct group . despite this seeming complexity in the distribution of byproduct operators , we prove in appendix [ sec : preparation ]
that the random outcomes @xmath95 of preparation measurements on @xmath4 lead the random quadratic and linear terms in the polynomial @xmath88 associated with @xmath22 to be uniformly random , simplifying our analysis . in the second stage of our protocol
we apply a final series of @xmath6 single - qubit pauli measurements to our output state which , while ideally equal to @xmath114 , will realistically be some mixed state @xmath70 .
the choice of single - qubit measurement bases depends on whether we are implementing sampling or verification , which can be chosen randomly with @xmath115 probability . during sampling
, we simply measure all qubits in the @xmath1 basis to generate a sample from the distribution @xmath116 , exactly as described in section [ sec : intractable ] .
although the randomness in the @xmath17 associated with @xmath70 means that we will almost certainly obtain each sample from a different distribution @xmath58 , our mqc sampling protocol remains classically intractable nonetheless . to prove this classical intractability
, we can treat the overall process of preparing a random @xmath70 and then sampling an outcome @xmath117 as itself a sampling process with probability @xmath118 .
given this description , and our knowledge of the complete randomness of the byproduct contributions @xmath42 , stockmeyer approximate counting can then be used to estimate @xmath89 as a conditional probability which is directly proportional to @xmath118 .
this suffices to proves theorem [ thm : one ] using the same arguments as in other classically intractable sampling proposals , the details of which are given in appendix [ sec : sampling ] .
if we choose to perform verification instead of sampling , then we measure all qubits in the @xmath3 basis , except for a random qubit @xmath99 which is measured in @xmath1 .
the outcome of this measurement @xmath119 is then fed into a parity function @xmath120 , where @xmath121 is defined in eq .
( [ eq : partial ] ) .
this process results in an output value of 0 or 1 , which we show in appendix [ sec : verification ] gives the same information as a measurement of the nonlocal stabilizer @xmath71 described in eq .
( [ eq : hami ] ) , with outcome @xmath122 .
because of our ability to characterize the closeness of @xmath70 to @xmath114 using measurements of @xmath71 , this means that we can interpret @xmath123 as a successful verification measurement , and @xmath124 as a deviation of @xmath70 from our intended @xmath22 . by obtaining many samples of @xmath125 for random @xmath99 ,
@xmath126 , and @xmath70 , the resultant estimate of @xmath127 lets us guarantee the classical intractability of our mqc sampling protocol to any desired statistical significance using only @xmath128 rounds of verification measurements , as stated in theorem [ thm : two ] .
[ thm : two ] suppose that the empirical average of our parity function after @xmath129 verification measurements satisfies @xmath130 , for the @xmath60 appearing in theorem [ thm : one ]
. then we can conclude with probability @xmath131 that our sampling distributions @xmath58 satisfy the assumptions of theorem [ thm : one ] , and thus generate classically intractable sampling .
we give a detailed proof of theorem [ thm : two ] in appendix [ sec : verification ] .
we should mention that another potential means of verifying the classical intractability of our sampling protocol would have been to directly measure the @xmath132 local stabilizers of our resource state @xmath4 , analogous to the technique used in @xcite .
the idea behind this verification scheme is that , if we guarantee our mqc resource state to be the ideal @xmath4 , then performing our prescribed pauli measurements should always generate the ideal sampling states @xmath22 .
unfortunately , this resource state verification scheme does nt detect errors occurring during preparation measurements , so that even when given an ideal mqc resource state , measurement imperfections during state preparation will still lead to logical errors which harm our output sampling state @xmath70 . in order for this verification scheme to rigorously guarantee the classical intractability of sampling in our setting , the single - qubit error rates for measurement must be less than @xmath133 , whereas our verification technique only needs errors rates of @xmath134 .
since this latter rate is the maximum allowed for any kind of sampling to maintain a constant variational error , this shows our verification scheme to be optimal with regards to its soundness under measurement imperfections .
we have demonstrated the use of mqc to perform classically intractable sampling and verification in a unified manner , with identical resource requirements for each task .
this shows that verifying the hardness of a quantum sampling protocol does nt need to be any harder than the actual sampling , and in certain architectures comes essentially for free .
this contrasts sharply with many existing quantum supremacy proposals@xcite , for which verifying the non - classical nature of sampling is significantly harder than the sampling itself , likely requiring exponential computational resources to ensure correctness . by using nonadaptive mqc to drive our protocol ,
we have furthermore allowed both sampling and verification to be carried out in constant time , which minimizes the effect of environmental decoherence , and potentially allows us to avoid the use of quantum error correction . as an outlook , we expect that a hybrid mqc sampling platform combining the simple physical implementation of @xcite or @xcite with the convenient theoretical analysis and flexibility available here would represent an extremely appealing framework for implementing classically intractable sampling .
in particular , a sampling protocol using nonadaptive mqc with non - clifford @xmath135 gadgets embedded in a 2d brickwork - type lattice could potentially demonstrate quantum supremacy in constant time using only @xmath136 qubits , and with entirely local interactions .
such a protocol would implement the sparse " iqp circuits appearing in @xcite , which require only @xmath136 two - body interactions . while this can be implemented in our framework using a 2d lattice of @xmath137 qubits which generalizes our @xmath4 , the possibility of reducing resource requirements further , potentially to @xmath136 qubits , would require using local complementation operations on graph states .
as these operations can quickly generate long - range entanglement using only local @xmath2 basis measurements , we consider such capabilities to represent a unique feature of mqc which are well - suited to reproducing the long - range , low - depth quantum circuits often utilized for quantum sampling .
this work was supported in part by national science foundation grants phy-1314955 , phy-1521016 , and phy-1620651 . 99 p.w .
shor , _ polynomial - time algorithms for prime factorization and discrete logarithms on a quantum computer _ , siam j. sci .
* 26 * , 1484 ( 1997 ) .
grover , _ quantum mechanics helps in searching for a needle in a haystack _ , phys .
lett . * 79 * , 325 ( 1997 ) .
d. deutsch and r. jozsa , _ rapid solution of problems by quantum computation _ ,
london a * 439 * , 553 ( 1992 ) .
terhal , d.p .
divincenzo , _ adaptive quantum computation , constant depth quantum circuits and arthur - merlin games _ , quant .
* 4 * , 134145 ( 2004 ) .
bremner , r. jozsa , d.j .
shepherd , _ classical simulation of commuting quantum computations implies collapse of the polynomial hierarchy _ , proc .
a * 467 * , 459472 ( 2011 ) .
s. aaronson and a. arkhipov , _ the computational complexity of linear optics _ , theory of computing , 9(4):143252 , ( 2013 ) .
t. morimae , k. fujii , and j.f .
fitzsimons , _ on the hardness of classically simulating the one clean qubit model _ , phys .
lett . * 112 * , 130502 ( 2014 ) .
bremner , a. montanaro , and d.j .
shepherd , _ average - case complexity versus approximate simulation of commuting quantum computations _ , phys .
* 117 * , 080501 ( 2016 ) .
s. rahimi - keshari , t.c .
ralph , c.m .
caves , _ sufficient conditions for efficient classical simulation of quantum optics _ , phys .
x * 6 * , 021039 ( 2016 ) .
e. farhi and a.w .
harrow , _ quantum supremacy through the quantum approximate optimization algorithm _ , arxiv:1602.07674 , ( 2016 ) .
rohde , d.w .
berry , k.r .
motes , j.p .
dowling , _ a quantum optics argument for the # p - hardness of a class of multidimensional integrals _ , arxiv:1607.04960 , ( 2016 ) .
_ , _ characterizing quantum supremacy in near - term devices _ , arxiv:1608.00263 , ( 2016 ) .
x. gao , s .-
wang , and l .- m .
duan , _ quantum supremacy for simulating a translation - invariant ising spin model _ , phys .
lett . * 118 * , 040502 ( 2017 ) .
k. fujii , _ noise threshold of quantum supremacy _ , arxiv:1610.03632 , ( 2016 ) .
m. bremner , a. montanaro , and d. shepherd , _ achieving quantum supremacy with sparse and noisy commuting quantum computations _ , arxiv:1610.01808 , ( 2016 ) .
b. fefferman , m. foss - feig , and a.v .
gorshkov , _ exact sampling hardness of ising spin models _ , arxiv:1701.03167 , ( 2017 ) .
f. shahandeh , a.p .
lund , t.c .
ralph , _ quantum correlations in nonlocal bosonsampling _ , arxiv:1702.02156 , ( 2017 ) .
lund , m.j .
bremner , t.c .
ralph , _ quantum sampling problems , bosonsampling and quantum supremacy _ , arxiv:1702.03061 , ( 2017 ) .
j. bermejo - vega , d. hangleiter , m. schwarz , r. raussendorf , and j. eisert , _
architectures for quantum simulation showing quantum supremacy _ , arxiv:1703.00466 , ( 2017 ) .
a. deshpande , b. fefferman , m. foss - feig , and a.v .
gorshkov , _ complexity of sampling as an order parameter _ , arxiv:1703.05332 , ( 2017 ) .
t. kapourniotis and a. datta , _ nonadaptive fault - tolerant verification of quantum supremacy with noise _ , arxiv:1703.09568 , ( 2017 ) .
_ , _ boson sampling on a photonic chip _ ,
science * 339 * , 798801 ( 2013 ) .
m. tillmann , b. daki , r. heilmann , s. nolte , a. szameit , and p. walther , _ experimental boson sampling _ , nature photon .
* 7 * , 540544 ( 2013 ) .
crespi , a. _ et al .
_ , _ integrated multimode interferometers with arbitrary designs for photonic boson sampling _ , nature photon .
* 7 * , 545549 ( 2013 ) .
broome , a. fedrizzi , s. rahimi - keshari , j. dove , s. aaronson , t. ralph , and a.g .
white , _ photonic boson sampling in a tunable circuit _
, science * 339 * , 794798 ( 2013 ) .
_ , _ multi - photon boson - sampling machines beating early classical computers _ ,
arxiv:1612.06956 , ( 2016 ) .
r. raussendorf and h.j .
briegel , _ a one - way quantum computer _ ,
lett . * 86 * , 5188 ( 2001 ) .
r. jozsa , _ an introduction to measurement based quantum computation _ , arxiv : quant - ph/0508124 , ( 2005 ) .
briegel , d.e .
browne , w. dr , r. raussendorf , and m. van den nest , _ measurement - based quantum computation _ , nature physics * 5 * , 1926 ( 2009 ) .
d. hangleiter , m. kliesch , m. schwarz , and j. eisert , _ direct certification of a class of quantum simulations _ , arxiv:1602.00703 , ( 2016 ) .
meyer and l.j .
stockmeyer , _ the equivalence problem for regular expressions with squaring requires exponential space _ , in proceedings of the 13th ieee symposium on switching and automata theory , pp .
125129 ( 1972 ) .
stockmeyer , _ the polynomial - time hierarchy _ , theor .
sci . * 3 * , 122 ( 1977 ) .
a. ehrenfeucht and m. karpinski , _ the computational complexity of ( xor , and)-counting problems _ , technical report 8543-cs , ( 1990 ) .
stockmeyer , _ on approximation algorithms for # p _ , siam j. comput . *
14 * , 849861 ( 1985 ) .
the proof actually allows for the existence of some multiplicative error , in the form of distributions @xmath138 which satisfy @xmath139 for all outcomes @xmath46 , with @xmath140 a fixed polynomial .
while sampling from such a distribution @xmath138 is still classically intractable , this is unsatisfactory from a practical standpoint .
for example , if any outcome @xmath141 satisfies @xmath142 , then we must have the probability @xmath143 be exactly 0 as well .
this is clearly impossible to verify for any experimental distribution @xmath138 , leading exact classically intractable sampling results to have a more strained relationship with experimental realities than their average - case counterparts .
d. gottesman , _ the heisenberg representation of quantum computers _
, talk at international conference on group theoretic methods in physics ( 1998 ) , arxiv : quant - ph/9807006 .
j. miller and a. miyake , _ hierarchy of universal entanglement in 2d measurement - based quantum computation _ , npj quantum information * 2 * , 16036 ( 2016 ) .
d. gottesman and i. l. chuang , _ demonstrating the viability of universal quantum computation using teleportation and single - qubit operations _ , nature * 402 * , 390 ( 1999 ) .
s. aaronson and l .- j .
chen , _ complexity - theoretic foundations of quantum supremacy experiments _ ,
arxiv:1612.05903 , ( 2016 ) .
m. rossi , m. huber , d. bru , and c. macchiavello , _ quantum hypergraph states _ , new j. phys . * 15 * , 113022 ( 2013 ) .
o. ghne , m. cuquet , f.e.s .
steinhoff , t. moroder , m. rossi , d. bru , b. kraus , and c. macchiavello , _ entanglement and nonclassical properties of hypergraph states _ , j. phys .
a * 47 * , 335303 ( 2014 ) . for comparison , the familiar complexity classes p and np
are respectively contained in the zero and first levels of the polynomial hierarchy ( ph ) .
while the randomized complexity class bpp has only been proven to lie in the second level of the ph , a proof of the widely conjectured p = bpp would place it in the zero ( lowest ) level as well . as a corollary , proving
p = bpp would allow stockmeyer counting to be implemented in the second level of the ph , causing the hypothetical collapse invoked in classically intractable sampling results to occur at the second level of the ph , rather than the third .
s. toda , _ pp is as hard as the polynomial - time hierarchy _
, siam j. comput . *
20 * , 865877 ( 1991 ) . on the other hand ,
the effect of the measurements used in our verification scheme on the measured state is different from that of direct measurements of @xmath71 .
for example , performing a genuine quantum nondemolition measurement of @xmath71 on the sampling state @xmath22 would leave it unchanged , whereas our scheme always collapses it to a tensor product of single - qubit @xmath1 and @xmath3 eigenstates .
since we only care about measurement statistics and not the post - measurement state , this has no impact on our protocol .
t. morimae , y. takeuchi , and m. hayashi , _ verified measurement - based quantum computing with hypergraph states _ , arxiv:1701.05688 , ( 2017 ) .
we now discuss the relationship of our work to previous proposals for classically intractable sampling with qubits , the class of boson sampling protocols having a largely different flavor with regards to theoretical underpinnings and experimental implementations .
as mentioned before , our work is most closely related to that of @xcite , as it implements their circuit - based iqp sampling in the context of mqc .
we have seen that this translation has several practical advantages , mainly that it allows us to use constant depth quantum circuits generated by local interactions to perform classically intractable sampling in constant time .
this translation also reveals the role of mqc byproduct operators in simplifying our protocol , with an associated randomness which ends up having no impact on the classical intractability of sampling .
furthermore , the convenient verification scheme utilized in our protocol can be applied equally well in any classically intractable sampling implementation using the iqp sampling states associated with conjecture [ conj ] , revealing an inherent practical advantage of sampling from this class of states .
this advantage more generally applies to any protocol which samples from output distributions defined by so - called hypergraph states@xcite .
although our work does nt make use of the alternate conjecture 2 of @xcite , concerning the average - case hardness of estimating fully - connected ising partition functions , our techniques can be easily generalized to define a similar mqc sampling protocol which relies upon conjecture 2 . in this alternate protocol
, our @xmath83 gadget would be replaced by gadgets for the non - clifford @xmath135 and @xmath144 gates , and our byproduct group would contain not only @xmath78 , but also @xmath145 gates . in terms of the clifford hierarchy of unitary operations @xcite ,
the pattern which emerges here is that using gadgets which implement operations at the third level of the clifford hierarchy leads to a random byproduct group formed from clifford gates at the second level of the clifford hierarchy .
just as with our protocol , this would eliminate the need to apply any clifford gates by hand " , reducing the physical resources needed for sampling .
our work also has many similarities to the mqc sampling protocol of @xcite , which similarly runs in constant time using a fixed brickwork " resource state preparable by a constant depth quantum circuit , and also allows for verification . in our protocol
, the average - case hardness of sampling relies on conjecture [ conj ] , while the average - case hardness in @xcite relies upon a conjecture regarding the estimation of output probabilities of random quantum circuits , argued in @xcite to be a stronger assumption . on the other hand
, this latter conjecture is very similar to that used in @xcite . on a different note ,
the simple byproduct group appearing in our protocol , which is necessary for our preparation measurements to always implement iqp circuits , allows us to carry out verification using only single - qubit measurements on our output sampling states @xmath70 .
in contrast , the more general unitaries implemented in @xcite would likely preclude any simple verification schemes based on the ideas of @xcite .
here we study the preparation stage of our mqc protocol , and show that the polynomials @xmath88 associated with our random output states @xmath70 contains uniformly random quadratic and linear coefficients , so that every @xmath34 and @xmath146 is an independent binary random variable with equal @xmath115 probability .
we show this by first characterizing the distribution of preparation outcomes @xmath147 , where @xmath93 , then using this to characterize the distribution @xmath148 of byproduct polynomials " arising in our protocol . we show that @xmath148 is uniformly random , a fact which holds true in the presence of arbitrary noise with spatial correlations of a bounded distance .
this result will be used in our proofs of sampling and verification in appendices [ sec : sampling ] and [ sec : verification ] .
we calculate @xmath147 using the born rule , which in our ideal setting says that given @xmath39-dependent preparation measurements on @xmath4 , the probability of obtaining an outcome @xmath149 ( where @xmath39 denotes the appropriate single - qubit eigenbases ) is @xmath150 the expression @xmath151 here denotes not a scalar , but a partial inner product on @xmath4 , consisting of an @xmath6-qubit state which is nt measured until the sampling and verification stage of our protocol .
consequently , eq . ( [ eq : t_prob ] ) says that @xmath147 is equal to the squared norm of this state @xmath151 .
although we would expect this output state to be the sampling state @xmath22 , a careful calculation of the inner products arising in our protocol reveals an additional @xmath152 scalar factor per preparation measurement , as remarked in figure [ fig : gadgets ] .
this shows that @xmath153 , where @xmath154 , which then proves the preparation measurement outcomes to be distributed as @xmath155 .
we note that this independence of measurement outcomes is a generic feature of mqc state preparation protocols , as the implementation of norm - preserving unitary operations in every preparation measurement will necessarily force eq .
( [ eq : t_prob ] ) to take a constant value for all @xmath95 , corresponding to every preparation outcome @xmath156 being uncorrelated and uniformly random .
we now use the uniform randomness of preparation measurement outcomes @xmath95 to prove the uniform randomness of byproduct polynomials @xmath42 , which depend on @xmath95 as @xmath157 .
these global byproducts arise from the local byproduct operators associated with random outcomes @xmath156 in each of the mqc gadgets shown in figure [ fig : gadgets ] , which are then commuted through our computation to contribute linear and quadratic terms to @xmath157 .
each quadratic and linear coefficient in @xmath42 can thus be expressed as a sum ( mod 2 ) of many different measurement outcomes @xmath156 , and it is clear that the complete randomness of each measurement outcome will lead every byproduct coefficient in @xmath42 which contains even a single random @xmath156 to be itself completely random .
it is clear that every quadratic coefficient contains contributions from at least one random @xmath156 , with the one exception of @xmath158 .
because our @xmath83 gadgets only apply @xmath83 byproduct operators between nearest neighbor logical qubits , and since qubits 1 and @xmath6 are never adjacent to each other in the circuit diagram of figure [ fig : protocol ] , it remains possible that @xmath158 will always be 0 .
a simple fix for this is to simply vary the ordering among each triple of qubits entering a non - clifford gadget using @xmath0 gadgets , so that all qubits are adjacent to all other qubits equally often . in this case , every quadratic coefficient @xmath159 in @xmath157 will receive @xmath160 random contributions from outcomes @xmath156 arising in @xmath83 gadgets , and every linear coefficient @xmath161 will receive @xmath97 contributions from outcomes arising in 1d cluster wires and @xmath0 gadgets .
this clearly proves that the distribution of byproduct operators will be uniformly random as @xmath162 , where @xmath163 . the above analysis which counts the number of measurement outcomes contributing to each coefficient of @xmath42 is unnecessary in an idealized setting , but is useful in the presence of realistic noise and experimental imperfections .
we can generally characterize this behavior as a trace preserving quantum operation @xmath164 which maps our mqc resource state to some imperfect @xmath165 .
our measurement statistics @xmath147 in this setting are again set by the born rule , but now as @xmath166 \\ & = { \mathrm{tr}}\left[{\ensuremath{|{\psi_\mathrm{prep}}\rangle\langle{\psi_\mathrm{prep}}| } } \mathcal{e}^\dagger({\ensuremath{|{\mathbf{t}}_{{\mathbf{a}}}\rangle\langle{\mathbf{t}}_{{\mathbf{a}}}|}})\right],\end{aligned}\ ] ] where @xmath167 represents the quantum operation which is adjoint to @xmath164 . while @xmath167 may modify our measurement projectors @xmath168 so as to displace or correlate the probabilities of local outcomes @xmath156 , we noted above that the coefficients of byproduct polynomials are determined by at least @xmath160 different such measurement outcomes , any one of which is capable of completely randomizing the probability of that coefficient .
consequently , in order for noise to alter the distribution of byproduct operators , the operator @xmath167 must induce correlations between at least @xmath160 different measurement outcomes in our system . in the presence of any noise with a finite correlation length , this is clearly impossible , which proves the uniform randomness of byproduct operators to be a robust property of our mqc protocol .
here we give a detailed proof of the classical intractability of our mqc sampling protocol under constant variational noise in the output sampling distributions @xmath58 .
we first discuss the general idea behind average - case classically intractable sampling protocols , so as to make clear what precisely needs to be demonstrated in our proof .
we then describe the use of classical post - processing on our measurement records to implement coarse - graining " in the description of our protocol .
this coarse - graining lets us simplify the analysis of failure probabilities required in our proof , and eventually lets us prove theorem [ thm : one ] , with its associated variational error threshold of @xmath169 .
we note a certain duality between the proof given here and the proof of theorem [ thm : two ] given in appendix [ sec : verification ] , with the former using a guaranteed bound on @xmath60 as a starting point and the latter deriving such a bound on @xmath60 as an end result .
any proof of classical intractability of quantum sampling requires adopting somewhat of a dual viewpoint . on the one hand
, we recognize that our sampling procedure is an intrinsically quantum task , but at the same time assume that the sampling distributions arising from this quantum process can be exactly replicated using a probabilistic classical algorithm .
this assumption , analogous to the assumption of a hidden variable model describing our quantum process , is made in order to derive a ( widely conjectured ) contradiction , the collapse of the polynomial hierarchy of complexity theory .
even though the probabilities of individual sampling outcomes @xmath89 are exponentially small and would require exponential time to estimate empirically , if they arise from a classical sampling process , then the technique of stockmeyer approximate counting can be used to estimate these probabilities up to multiplicative error . in particular ,
stockmeyer counting can be used to output an estimate @xmath170 which is related to our probability of interest by @xmath171 , for any desired polynomial @xmath140 .
the use of an average - case complexity conjecture , like conjecture [ conj ] in our paper , is then required to connect the ability to estimate such probabilities in the presence of noise to the ability to solve # p - hard problems , from which a collapse of the polynomial hierarchy follows .
stockmeyer counting is an unphysical process which can not be carried out efficiently using classical or quantum devices , but can be implemented with a hypothetical alternating turing machine " capable of efficiently solving problems in the third level of the polynomial hierarchy @xcite .
furthermore , stockmeyer counting involves manipulations on a register of binary random variables underlying our random outcomes , and consequently can only estimate probabilities arising as outcomes of classical randomized algorithms .
nonetheless , if we assume the existence of an efficient classical algorithm for exactly sampling from the distribution @xmath172 , stockmeyer sampling would then permit a device existing in the third level of the polynomial hierarchy to estimate any @xmath50 up to multiplicative error , and thus solve any problem in # p. because solving arbitrary problems in # p is known by toda s theorem @xcite to allow one to efficiently solve all problems in the hierarchy , assuming the existence of this efficient classical algorithm for sampling from distributions @xmath59 would necessarily collapse the polynomial hierarchy to its third level , a contradiction .
hence , this proves the task of sampling from arbitrary @xmath59 to be classically intractable . a necessary ingredient in any _ average - case _ classically intractable sampling result is a mathematical problem whose estimation remains # p - hard even when our estimates have some finite probability of failing to be multiplicatively close to their actual value .
in our setting , this problem is furnished by conjecture [ conj ] , which says that estimating @xmath50 up to @xmath54 multiplicative error is # p - hard , even when a fraction @xmath173 of our estimates fail to lie within this @xmath54 multiplicative bound . evidence in support of conjecture [ conj ]
is given in @xcite .
this failure probability @xmath64 ends up determining the allowed deviation of our quantum sampling distributions @xmath58 from their ideal @xmath59 .
if this deviation is sufficiently small , as measured by the variational distance between @xmath58 and @xmath59 , the assumed computational hardness of estimating @xmath50 then guarantees that our quantum sampling task will be classically intractable .
consequently , our main goal in this proof is to analyze the deviations in our distributions @xmath116 arising from deviations in our experimental states @xmath70 from their ideal @xmath114 , and to find sufficient conditions to guarantee that the failure probability in estimating @xmath50 using stockmeyer sampling on @xmath58 is below our threshold @xmath64 .
we now introduce the idea of coarse - grained sampling distributions , which indeed we have already implicitly made use of in the description of our sampling protocol . in section [ sec :
protocol ] , we described different preparation outcomes @xmath93 as giving rise to different ideal sampling states @xmath94 via the correspondence @xmath96 .
this means that whenever different preparation outcomes @xmath174 generate the same byproduct polynomials @xmath175 , the resultant sampling states will be identical . in reality though , it is entirely possible that these preparation outcomes will generate different sampling states @xmath176 , leading our description of a single sampling state @xmath177 to represent a coarse - graining over equivalent preparation outcomes @xmath95 .
in particular , if @xmath147 denotes the probability of obtaining a preparation outcome @xmath95 arising from our @xmath39-dependent pauli measurements on @xmath4 , then we find @xmath70 to be given by @xmath178 @xmath179 represents a normalization factor which gives the total probability on input @xmath39 of obtaining any outcome @xmath95 associated with the byproduct polynomial @xmath180 .
while the above coarse - graining might appear trivial , we will now show how this can be used to effectively mix the inequivalent states @xmath70 and @xmath181 when @xmath17 and @xmath182 differ only in their linear coefficients .
if we describe our overall sampling process at this stage as first preparing a random state @xmath70 with @xmath183 , which is then sampled to obtain an @xmath1 basis outcome of @xmath117 , then we would record this in an experiment as yielding the outcome @xmath184 in some outcome space @xmath185 . from the layout of our sampling protocol ,
the probability of this outcome is clearly @xmath186 . because of the degeneracy @xmath187 for all outcomes @xmath46
, we say that any such outcome samples from the polynomial @xmath17 .
these exponentially many outcomes are precisely the ones which can be used to obtain an estimate of @xmath50 via stockmeyer counting , and we will choose our coarse - graining to eliminate this degeneracy , so that each @xmath50 is determined by a unique sampling outcome from a unique output sampling state .
we note that this coarse - graining was used implicitly in @xcite , although interpreted there as an obfuscation " of output probabilities . in appendix
[ sec : preparation ] we showed that the distribution of byproduct polynomials is uniformly random as @xmath188 , where @xmath163 . given this robust characterization of @xmath179
, we will use @xmath189 to indicate the conditional probability of obtaining any outcome which samples from @xmath190 , given that the quadratic portion of our byproduct polynomial is @xmath40 .
this leads @xmath189 to be @xmath191 we use @xmath192 to indicate the @xmath1 basis outcome string corresponding to the linear terms of @xmath17 . in the above
, we have also defined @xmath193 to be the state @xmath194 where @xmath195 indicates a product of @xmath3 operators . in the ideal
setting where each @xmath72 , the result of applying @xmath196 to @xmath197 is to simply remove the linear components of @xmath17 , leaving the state @xmath198 . in this idealized
setting , the result of averaging over all @xmath197 and applying the correction @xmath196 in each case is to leave the state @xmath199 , which contains only cubic and quadratic terms . while we ca nt literally implement these unitary corrections @xmath196 within the setting of mqc
, we can simulate their action through classical postprocessing on our measurement outcomes .
in particular , whenever we obtain an outcome of @xmath184 in our sampling experiment , we instead record this as a coarse - grained outcome @xmath200 lying in a simpler outcome space @xmath201 .
this is equivalent to recording only the polynomial @xmath17 sampled by our experiment , and forgetting the relative contributions to @xmath17 from mqc byproduct operators and from sampling outcomes @xmath46 .
the equivalence of this coarse - graining in our measurement records with the action of active unitary corrections arises from the equality @xmath202 used to derive eq .
( [ eq : prob_1 ] ) . given this coarse - grained description of our experiment
, we would like to bound the failure probability @xmath203 of obtaining an estimate @xmath66 which differs from the true @xmath50 by more than a multiplicative factor of @xmath54 . by requiring this probability to be less than the @xmath204 appearing in conjecture [ conj ] , we will arrive at concrete conditions on our coarse - grained output states @xmath193 in order for our mqc protocol to implement classically intractable sampling .
while the stockmeyer counting used to obtain @xmath66 from our sampling probabilities @xmath189 technically introduces its own multiplicative error in this estimate , because this error can be reduced in our ( hypothetical ) stockmeyer counting algorithm to any inverse polynomial @xmath205 while still retaining a polynomial runtime , we will ignore this error in the following and simply set @xmath206 .
we first use markov s inequality to bound the probability of our estimate @xmath66 failing to lie within an arbitrary constant distance of @xmath50 , @xmath207 , over arbitrary polynomials @xmath190 .
we will later convert this into a failure probability for obtaining an estimate of @xmath50 outside of our allowed @xmath54 multiplicative error . since the approximate and exact values of @xmath50 can both be interpreted as probabilities in different distributions , @xmath206 and @xmath208
, we find that the distance between these values , averaged over @xmath41 with fixed @xmath43 , is proportional to the variational distance between these distributions as @xmath209 defining @xmath210 to be the variational distance between these distributions , markov s inequality then tells us that for any @xmath211 and for @xmath190 with a fixed @xmath43 , @xmath212 having this bound in hand , we now give an anticoncentration bound on the probability that @xmath213 , which lets us convert the above bound into a statement about the failure probability @xmath203 .
we utilize a particular form of cantelli s inequality stating that for any non - negative random variable @xmath1 and constant @xmath214 in @xmath215 , @xmath216 this agrees with the more well - known paley - zygmund inequality at @xmath217 , but otherwise gives a more stringent upper bound .
setting @xmath218 , @xmath219 , and using the result @xmath220 from @xcite , this lets us restrict the probability of @xmath221 being less than @xmath222 as @xmath223 we now define @xmath224 to be the average variational distance between distributions @xmath225 and @xmath226 , averaged over all @xmath43 . combining eq .
( [ eq : markov ] ) with the average of eq .
( [ eq : pz ] ) over all @xmath39 , this results in a bound on the multiplicative failure probability of @xmath227 which holds for every @xmath228 .
we now require the failure probability to be at most @xmath229 , in line with conjecture [ conj ] , and numerically optimize over @xmath230 to find the largest allowed value of @xmath60 for which this can be achieved .
this yields a maximum of @xmath231 , which has a rational lower bound of @xmath232 .
this completes our proof of theorem [ thm : one ] .
here we prove that the verification scheme occurring in the last stage of our mqc protocol does indeed guarantee the classically intractable of our sampling process .
we first show that the local @xmath1 and @xmath3 measurements made on our sampling states @xmath70 during verification correspond to exact measurements of the nonlocal stabilizers @xmath71 , via the parity functions @xmath125 .
this allows us to estimate the average @xmath233 with respect to random @xmath70 , which allows us to bound the average variational distance @xmath234 using results from @xcite .
if our empirical estimate of @xmath233 remains sufficiently low , an application of hffding s inequality lets us show that @xmath128 verification measurements are sufficient to conclude that @xmath235 with any fixed statistical significance , proving theorem [ thm : two ] .
we first briefly review our verification procedure .
after preparation of a random @xmath70 , we choose with 50% probability to perform either sampling or verification measurements on @xmath70 .
if verification is chosen , we further choose a random qubit @xmath99 of @xmath70 which is measured in @xmath1 , while all other @xmath236 qubits are measured in @xmath3 .
we denote the measurement outcome string by @xmath119 , ignoring the fact that @xmath237 is associated with a different measurement basis .
we then use our knowledge of the polynomial @xmath17 associated with @xmath70 to compute a parity function of @xmath126 , @xmath120 , where @xmath238 is the polynomial difference @xmath239 .
it is easy to show that @xmath121 is independent of the value of @xmath237 .
we show here that the process of measuring @xmath126 using single - qubit pauli measurements and then computing @xmath125 is exactly equivalent to measuring the nonlocal stabilizer @xmath71 as @xmath240 , where @xmath241 indicates the @xmath71 outcome corresponding to @xmath126 .
both processes yield binary random variables as their output , and in order to prove that their probability distributions are identical , we can prove that both measurement schemes are associated with identical hermitian observables .
while measurements of @xmath71 are clearly associated with the hermitian operator @xmath71 itself , it is nt immediately clear how we should interpret the measurements of @xmath126 as measuring any particular hermitian operator .
the answer comes by recognizing that our relevant measurement statistics during verification consist only of the binary values @xmath125 , and forgets the specific outcomes @xmath126 which produced them .
translating these @xmath242 outcomes into equivalent @xmath71 outcomes shows the expectation value of @xmath71 on @xmath70 to be @xmath243 \\ & = { \mathrm{tr}}\left[\rho_f \left ( x_i \sum_{{\mathbf{v } } } ( -1)^{\partial_{i}f({\mathbf{v } } ) } { \ensuremath{|{\mathbf{v}}\rangle\langle{\mathbf{v}}|}}\right)\right ] \\ & = { \mathrm{tr}}\left(\rho_f { h_f^{(i)}}\right).\end{aligned}\ ] ] in the last equality , we have used the definition of @xmath71 in eq .
( [ eq : hami ] ) , while in the second to last equality we used @xmath244 .
this reveals that the expectation value of @xmath122 is equal to that of @xmath71 on @xmath70 , and since we made no assumptions about @xmath70 , this shows that our verification scheme is exactly equivalent to measuring @xmath71 @xcite . as a concrete example , suppose we are working with the 3-qubit sampling state @xmath245 and wish to measure the stabilizer @xmath246 . in this case
, we would perform our verification by measuring @xmath1 on qubit 1 , @xmath3 on qubits 2 and 3 , and then computing the polynomial @xmath247 .
this process , which can be thought of as obtaining classical values and plugging them in to the stabilizer itself , would indicate a success when @xmath248 and @xmath249 , or when @xmath250 and at least one of @xmath251 or @xmath252 holds true . given the ability to measure arbitrary @xmath71 using single - qubit @xmath1 and @xmath3 measurements
, we now note that the average @xmath253 over randomly chosen sites @xmath99 is equal to 1 on a given @xmath70 only when @xmath72 is the ideal sampling state .
more generally , the techniques of @xcite show that this average can be used to bound the closeness of @xmath70 to @xmath114 , as measured by the fidelity @xmath254 . for our purposes , it will be more convenient to work with the square of this quantity , @xmath255 . when @xmath256 , @xmath70 can not be orthogonal to @xmath22 , and must have a fidelity squared of at least @xmath257 .
if we average both sides of this equality over polynomials @xmath88 with random @xmath42 , then we find that the average fidelity squared @xmath258 of output states @xmath70 relative to their intended @xmath22 is bounded by the average @xmath233 as @xmath259 with eq .
( [ eq : fidelity_bound ] ) in hand , we can now bound the average variational distance @xmath234 between the sampling distributions arising from @xmath70 and @xmath22 .
we utilize the fact that the quantum 1-norm distance @xmath260 gives an upper bound on the variational distance of any output sampling distributions , where @xmath261 with @xmath262 the operator absolute value .
we also use a well - known bound on the 1-norm distance , @xmath263 , which together yield @xmath264 in the above , we used the two bounds mentioned , as well as jensen s inequality for the concave function @xmath265 in eq .
( [ eq : bound_two ] ) . using the relationship between the average of stabilizers and parity functions , @xmath266 , this
finally lets us show that in order to verify that @xmath267 , it is sufficient for our parity function average to be below @xmath268 this gives the bound appearing in theorem [ thm : two ] .
although any empirical estimate of @xmath233 obtained from finitely many measurements of @xmath125 is nt guaranteed to accurately reflect its true value , we can bound the closeness of this estimate with high probability using the uniformly random distribution of byproduct operators proved in appendix [ sec : preparation ] .
in particular , this tells us that for any fixed @xmath39 , the average @xmath269 over output random byproducts is unbiased towards any fixed @xmath70 , and thus is an accurate indicator of the uniform closeness of sampling states .
this lets us treat @xmath233 as a simple binary random variable , and use hffding s inequality to bound the probability of this estimate deviating too far from the true value of @xmath233 . hffding s inequality says that if we obtain an estimate @xmath270 of a binary random variable @xmath1 using @xmath271 independent samples , the probability of the true average @xmath272 lying above @xmath270 by more than @xmath273 is @xmath274 in our case , we choose @xmath1 to be our random parity function , and @xmath273 to be the difference between our specified tolerance @xmath275 , and the more numerically precise tolerance for classically intractable sampling derived in appendix [ sec : sampling ] , @xmath276 . setting @xmath277 , this gives a failure probability of @xmath278 converting this into a success probability @xmath279 then completes our proof of theorem [ thm : two ] .
a final remark is given to our means of measuring the highly nonlocal , non - pauli stabilizers @xmath71 through single - qubit pauli measurements .
this technique can actually be generalized to measure the stabilizers of any sampling state formed by starting with @xmath20 and applying an iqp circuit composed of @xmath145 , @xmath3 , @xmath78 , @xmath83 , and any higher multiply - controlled @xmath3 gates . as these states include all hypergraph states as special instances ,
our verification scheme consequently extends that of @xcite , which requires the measured hypergraph stabilizers to be supported on a constant number of qubits .
generalizing yet further , we see that the necessary and sufficient condition for a local measurement scheme to exactly replicate measurements of a nonlocal operator @xmath280 in this manner is that @xmath280 can be diagonalized in a basis which is a tensor product of single - qubit eigenbases .
while this allows us to measure many different multi - qubit operators using only single - qubit measurements , a simple counterexample is given by the hermitian operator @xmath0 , which can not be measured in this manner owing to its unique @xmath281 eigenstate being the entangled @xmath282 . | while quantum speed - up in solving certain decision problems by a fault - tolerant universal quantum computer has been promised , a timely research interest includes how far one can reduce the resource requirement to demonstrate a provable advantage in quantum devices without demanding quantum error correction , which is crucial for prolonging the coherence time of qubits .
we propose a model device made of locally - interacting multiple qubits , designed such that simultaneous single - qubit measurements on it can output probability distributions whose average - case sampling is classically intractable , under similar assumptions as the sampling of non - interacting bosons and instantaneous quantum circuits .
notably , in contrast to these previous unitary - based realizations , our measurement - based implementation has two novel features .
( i ) verifying the classical intractability of our sampling is done using the same resource requirements as sampling itself , requiring only a change in the pauli measurements bases .
( ii ) our implementation involves no adaptation of measurement bases , leading output probability distributions to be generated in constant time , independent of the system size .
thus , it could be implemented in principle without quantum error correction . |
ideas for classifying quantum phases of matter that go beyond the description due to landaus s symmetry breaking scheme were conceived since the discovery of the integer and fractional quantum hall effect and high temperature superconductivity @xcite . since
then the study of low temperature phases exhibiting topological order have been given new life due to their potential applications in fault tolerant computation @xcite and in their description of topological insulators and their generalizations in the form of symmetry protected topological phases @xcite .
these phases are usually described by a topological quantum field theory in the continuum .
there are also examples of exactly solvable hamiltonians describing the topological phases . in most cases
these hamiltonians are sums of commuting projectors , with the most prominent set of models being given by the quantum double models ( qdm ) of kitaev @xcite .
the qdm models on lattices with boundaries have been studied in the past @xcite .
the types of boundaries considered were either smooth or rough .
there are good reasons to consider models with boundary . in 3d , for example , it is not realistic to consider boundary less systems .
more importantly , the relationship between boundary and bulk degrees of freedom provide us new interesting phenomena .
these models are described with an exactly solvable hamiltonian which is split into bulk and boundary parts .
they exhibit different topological phases in the bulk and the boundary . in general the phase on the boundary
is characterized by anyon types belonging to a subcategory of the fusion category describing the bulk phase . in @xcite domain walls
are considered separating two different topologically ordered phases governed by the irr s of the quantum double of two different groups @xmath5 and @xmath6 .
the conditions for anyons to tunnel between these two phases are studied in detail . in the special case
when one of the groups become trivial the domain wall turns into a boundary , which have been analyzed in more detail in @xcite for the string - net models @xcite as well .
the set of anyons on the boundary are then determined by fusing the bulk anyons into the domain wall or in other words by _ condensing _ the bulk anyons on the boundary . in all these examples
the boundary is a physical trait of the lattice where the system lives on .
we may say that it is put in by hand and its location on the lattice is fixed . in this work we introduce exactly solvable models of the quantum double type that contain these domain walls as part of their spectrum .
this means that there are operators that act on the hilbert space of the theory that can create these domain walls locally as an eigenstate of the hamiltonian .
the reason this happens in these models is because the degrees of freedom on the local hilbert space are no longer labeled by elements of a group as it is in the case of the qdm of kitaev based on group algebras . here
the basis of the local hilbert space is generated by the same elements as in the group algebra plus an extra label .
in other words the model now has two degrees of freedom , namely the elements of the group algebra and an extra quantum number called @xmath7 belonging the to the set @xmath8 .
this introduces new excitations in the theory that can not be moved around like the deconfined anyonic excitations of the qdm of kitaev .
thus we can now create `` walls '' or `` regions '' of these excitations and we see that they are similar to the domain wall excitations studied in the past .
a more physical way of looking at this is that the model considered here has both a global and local @xmath0 symmetry .
thus their ground states obtain a degeneracy from two mechanisms namely the global symmetry breaking mechanism , as it happens in say the ising model and a topological degeneracy , when the system is placed on a surface with a non - zero genus as it happens in the toric code .
thus the global @xmath0 symmetry is broken at the ground sate level with the system settling down in either of the two toric code topological phases labeled by @xmath9 .. the domain wall states are obtained by operating in the sector controlled by the global symmetry , namely the @xmath7 degrees of freedom , and thus they separate regions that are locally in one of the two toric code phases .
thus we can interpret this system as a toric code decorated with an additional @xmath0 global symmetry whereas the toric code in it s original formulation has only a local @xmath0 symmetry .
recently such gauge systems decorated with a global symmetry have been studied under the name symmetry enriched topological ( set ) phase @xcite .
the ground states in these systems preserve the global symmetry unlike the examples studied here .
other examples with both a global and local symmetry include the quantum hall ferromagnets @xcite . these domain wall states can manifest themselves in different forms each with interesting properties .
the deconfined anyons on each of the phases can be fused into the domain walls to become a part of it .
this has been termed as fusing anyons into domain wall excitations in the literature .
once we fuse anyons into the domain wall we create states which are essentially isolated excitations connected to the domain wall through a string which then act like sinks for the other anyons that condense into them just like how they condense into the physical boundaries considered in @xcite .
thus these are sinks which can be created locally in any part of the lattice .
apart from this the domain walls can act as annihilators by killing states that fuse into them .
this is a new feature in this model .
finally they act as a scatterer helping us tunnel between the anyons of the two topologically ordered phases .
in particular they can switch between same anyon types or different anyon types .
the effect is similar to the anyon symmetry of bi - layer systems .
the paper is organized as follows .
section 2 describes the setup of our system which includes the hilbert space and the operators that go into the quantum double hamiltonian .
the spectrum is similar to the that of the qdm .
the hamiltonian is a sum of commuting projectors and thus all its eigenstates can be easily obtained .
the ground states and the anyonic excitations are presented in sections 3 and 4 respectively .
the domain wall states are also shown in section 4 .
their effects are studied in detail in section 5 . in section 6
we briefly generalize the model to the case where the two different topologically ordered phases are given by the qdm of a pair of distinct groups .
some crucial remarks make up section 7 .
the model is defined on a two dimensional square lattice with the local hilbert space @xmath10 on the links @xmath11 .
the degrees of freedom have an extra quantum number @xmath7 beyond the usual @xmath0 group elements @xmath12 and @xmath13 of the toric code .
this extra quantum number takes values in a discrete set with two elements , @xmath14 . thus a basis for the local hilbert space
is given by @xmath15 can be thought of as @xmath16 , where @xmath17 is the group algebra of @xmath0 and @xmath18 is a two dimensional vector space whose basis elements are labeled by the elements of @xmath14 .
thus the basis elements @xmath19 of @xmath10 can also be represented by @xmath20 . ] .
the full hilbert space is given by @xmath21 .
a basis of @xmath22 is given by @xmath23 the operators acting on the hilbert space @xmath22 can be written out of following local operators ^:=^^,where @xmath24 and @xmath25 are the pauli matrices which are eigenvectors of @xmath26 . ] with @xmath27 .
we define two very important projectors which we will call @xmath28 with @xmath29 .
their action is given by p^rs , g=(r , s)s , g .these projectors will play a very important role in this model as we will see later and they can be easily be written as linear combination of @xmath30
. the hamiltonian is made of vertex and plaquette operators like the toric code hamiltonian @xmath31 their action are however sightly different .
consider the vertex operator @xmath32 .
it is convenient to write the vertex operator as a sum of two other vertex operators @xmath33 .
the operator @xmath34 acts exactly like the usual toric code vertex operator flipping @xmath35 when all the local states around the vertex @xmath36 has quantum number @xmath37 .
otherwise , if at least one local state around the vertex @xmath36 has quantum number @xmath38 , the vertex operator @xmath34 projects the state to the zero vector of @xmath22 .
the action is shown in figure [ avphi ] .
the operator @xmath39 does the same for @xmath38 .
these operators can be realized as a_v^r = _
l(v)_l^x p_l^r ri , where @xmath40 . it is not difficult to see that the vertex operators obey the following relations for all vertices @xmath36 and @xmath41 @xmath42&= & 0 \;.\end{aligned}\ ] ] the plaquette operator @xmath43 is defined in a similar way , it is also made up of two others plaquette operators @xmath44 , which are defined as follows [ bp1 ] b_p^r = _
lp_l^z p_l^r ri , where @xmath45 . again , in the case where all the local states living around the plaquette have quantum number @xmath37 ( or @xmath38 ) the plaquette operator @xmath46 ( or @xmath47 ) gives the same state back with the eigenvalue equal to the holonomy around the plaquette @xmath48 . and if there is a mix of two different values of @xmath7 , the plaquette operator projects the state to the zero vector .
the plaquette operator obey the following relations for all plaquettes @xmath48 and @xmath49 @xmath50&= & 0 \;.\end{aligned}\ ] ] besides , the vertex and plaquette operators satisfy the following relations for all vertices and plaquettes @xmath51 & = & 0\;,\ ; \forall r , s\ ; \\
\label{ab - ba}\left[a_v , b_p\right ] & = & 0\;.\end{aligned}\ ] ] in other words the hamiltonian is a made up of a sum of commuting operators , which make it exactly solvable .
since the hamiltonian is a sum of commuting operators it can be easily diagonalized and just as in the toric code case , this model has degenerate ground states .
the vertex and plaquette operators obey the properties and which means their eigenvalues are either @xmath13 , @xmath52 or @xmath12 . due to this fact and equation
the vacuum subspace @xmath53 is defined as @xmath54 consider only states with @xmath37 on all the links , which we will represented by @xmath55 .
once restricted to this subspace of @xmath22 , the hamiltonian is exactly like the toric code hamiltonian , and one ground state is the following @xmath56 using the relations and it is not difficult to see that the above state satisfies @xmath57 .
there are three more ground states one can obtain from @xmath58 by flipping @xmath35 along a non contractible loop of the torus , in other words , by applying the operator @xmath59 where @xmath60 is a non contractible loop on the dual lattice of a torus .
since there are only two non contractible loops on the torus ( up to homotopic deformations ) the four ground states one can obtain are the ones represented in figure [ tcgroundstates - a ] . likewise , if we do the same analysis for the subspace of @xmath22 made up of only linear combinations of vectors of the kind @xmath61 , with @xmath38 on all the links , we will find out four more states which are also ground states of the hamiltonian .
these states are represented in figure [ tcgroundstates - c ] .
an important thing is that the states with @xmath37 can not be mapped into the ones with @xmath38 by gauge transformations ( vertex operator action ) , thus making the ground state degeneracy for this model twice the toric code ground state degeneracy .
this can also been seen due to the phenomena of spontaneous symmetry breaking of the global @xmath0 symmetry . a basis for the subspace @xmath53 is the following : @xmath62 .
the excitations of these model are similar to the toric code but with important differences according to the values of @xmath7 .
the elementary excitations are charges and fluxes but they now come in two kinds according to the @xmath7 label they carry .
the operators which creates excitations are string operators which act on either a path @xmath60 on the direct lattice or a path @xmath63 on the dual lattice . in
the following let us study each of these operators separately .
consider the string operator @xmath64 the @xmath65 operator trivially commutes with the plaquette operator and since @xmath66 , the state @xmath67 , with @xmath68 a ground state of the hamiltonian , is an excited state which has two charge excitations on the vertices @xmath69 and @xmath70 , as shown in figure [ avexc ] .
let us call these charge excitations @xmath71 , note they can be of two different kinds , namely @xmath72 and @xmath73 , depending on the @xmath7 value of the ground state it was created from .
the operator @xmath74 is the string operator which creates fluxes .
it trivially commutes with the vertex operators creating flux excitations on plaquettes @xmath75 and @xmath76 at the end points of @xmath63 , as shown in figure [ bpexc ] .
again , the flux excitations can be of two different kinds , namely @xmath77 and @xmath78 distinguished by the @xmath7 value .
all these excitations are toric code - like anyons , apart from the fact that the charges and fluxes are now of two different kinds .
since each kind of flux ( charge ) are created from different ground states , they can not be fused , in other words an anyon @xmath79 ( with @xmath80 and @xmath81 ) can only be fused with another anyon @xmath82 if @xmath83 , which means there are two distinct set of anyons whose generators are @xmath84 and @xmath85 .
the fusion rules of such anyons are the following e_sm_s & = & _ s , + e_s_s & = & m_s , + m_s_s & = & e_s , + e_s e_s & = & m_sm_s = _ s _ s = 1_s , where @xmath86 is called a dyon and @xmath87 stands for the vacuum with label @xmath7
. as we have seen before the plaquette and vertex operators have zero as an eigenvalue apart from the usual @xmath88 . a state which gives eigenvalue zero for a plaquette ( vertex ) operator
is said to have a zero flux ( charge ) excitation localized on such a plaquette ( vertex ) .
this kind of state can be obtained by flipping the quantum number @xmath7 of a single link @xmath11 belonging to @xmath89 ( or star(@xmath36 ) ) .
an operator that does this is given by @xmath90 .
at this point a graphical notation would be useful for understanding these excitations .
for a given local state @xmath91 we associate a diamond as shown in figure [ grafical ] .
these diamonds represent the @xmath7 quantum number . if @xmath37 we color it white as shown in figure [ grafical - b ] or if @xmath38 we color it green as shown in figure [ grafical - c ] .
the operator @xmath92 can be thought as the operator which flips the diamond color .
the properties of such an operator are the following @xmath93 & = & \left[f_l,\sigma^z \right ] = 0\;,\end{aligned}\ ] ] where @xmath94 .
let us now see how to create the zero fluxes excitations .
we act on the ground state , @xmath58 written in the previous section . as we already saw @xmath95 for all vertices and plaquettes .
the state @xmath96 has now a zero flux excitation localized on each plaquette which share the link @xmath11 , moreover it has also zero charge excitations localized on the vertices which are on the end points of link @xmath11 .
the reason for that is that the zero eigenvalues for the plaquette operators comes from the case where there are at least two diamonds of different colors sharing the same plaquette , but since the ground state @xmath58 has @xmath37 for all the links the plaquette operators wo nt give zero as eigenvalue .
however when the operator @xmath92 is applied on the link @xmath11 , it creates a single link of different diamond color which leads to these zero fluxes excitations on all the plaquettes that share the link @xmath11 .
likewise , the vertex operators on the end points of the link @xmath11 will also be excited as @xmath92 flips @xmath34 to an operator of the type shown in figure [ avphi - b ] .
this operator measures the zero charge excitations .
thus the state ends up with two zero flux excitations and two zero charges excitations .
thus we have @xmath97 the picture in figure [ domain - a ] illustrates this fact , showing the mismatch of diamond colors .
we call the zero flux excitation @xmath98 while the zero charge excitations are called @xmath99 .
if the flip operator @xmath92 is applied on the ground state @xmath100 along a path @xmath60 , that is @xmath101 it will create a couple of zero flux and zero charges all along it s path and not just in the end points like in the toric code .
we say that these excitations are not deconfined like the ones in the toric code .
this is shown in figure [ domain - b ] .
we can also apply the operator @xmath92 in all the links which are inside an area @xmath102 ( including its boundary @xmath103 ) .
this is done through applying the following operator @xmath104 in figure [ domain - c ] one can see the effect of applying such an operator on the ground state . note that there is no diamond color mismatch inside the area .
the zero flux and zero charge excitations will appear only on its boundary as it can be seen in figure [ domain - c ] .
the energy cost to create such a state is proportional to length of the path @xmath103 .
this state is a domain wall which separates the two different toric code sectors labeled by @xmath7 . in the case of figure [ domain - c ] , the state with @xmath37 outside the area and @xmath38 inside it .
these domain walls are very important , for example they allow toric code - like excitations to tunnel between two different values of @xmath7 apart from other effects .
these will be illustrated in what follows .
as we have just seen , the operator @xmath105 creates a domain wall @xmath103 which separates the two toric code sectors with different values of @xmath7 when it is applied on the ground state ( take the ground state @xmath58 for example ) .
the state with a domain wall is no longer a ground state since it has now zero flux and charge excitations .
however it is still an eigenstate of the hamiltonian with some eigenvalue @xmath106 , which is proportional to the length of the path @xmath103 .
now , toric code - like anyons can be created from this state which can be thought of as creating toric code - like excitations in the background of these domain walls .
consider the state @xmath107 drawn in figure [ domaintc - a ] .
a pair of anyons can be created at the end points of a path @xmath60 by the application of a string operator along this path .
let us consider two distinct cases : the case where @xmath108 and @xmath109 . in the first case
the path @xmath60 must be either entirely inside or outside the area @xmath102 , which means it will create a pair of charges either inside or outside the area @xmath102 , but as we already know the quantum number @xmath7 is different in these two regions . thus the anyonic excitations created outside will be of the kind @xmath110 while the ones created inside will be of the kind @xmath111 ( with @xmath112 ) .
the figures [ domaintc - b ] and [ domaintc - c ] illustrate this fact .
however interesting phenomena show up when the path crosses the boundary @xmath103
. we will discuss these next .
if there exist a pair of anyons outside of the area @xmath102 , these two anyons can be freely moved and separated from each other without any energy cost .
however if one of the anyons fuses with the zero excitations on the boundary , @xmath103 , it forms a zero excitation on the domain wall .
it forms a configuration with one unit of energy less .
note that this is not condensation as in the usual case of physical boundaries where the deconfined excitations of the bulk go to the ground state of the boundary hamiltonian . in the present case
we create an isolated flux along with the domain wall . as an example consider the flux excitations shown in figure [ cond - a ] .
when one of these fluxes is moved to the domain wall and fused with the zero flux excitation as shown in figure [ cond - b ] it will become a zero flux excitation as well .
thus we get the following fusion rule between the flux @xmath113 ( @xmath80 ) and a zero flux @xmath98 as [ condflux]m_0m_s = m_0 , si .
a similar thing happens with the deconfined charge excitations , which can be summarized by the following fusion rule [ condcharge]e_0e_s = e_0 , si . in the language of states in the theory , we begin with a broken ground state , @xmath100 and apply the operator @xmath105 resulting in the state @xmath114
this state has an energy @xmath106 proportional to the perimeter of the area @xmath102 .
the square lattice is now divided into the toric code with @xmath37 outside the region @xmath102 and the toric code with @xmath38 inside the region @xmath102 .
the boundary is the energetic domain wall separating the two sectors .
now apply a string operator creating fluxes denoted by @xmath115 as before .
this operator creates a pair of fluxes at the endpoints of @xmath63 in either sector .
let us create it in sector 1 and make the dual string , @xmath63 end at the domain wall .
this is the condition for fusing one of the fluxes , @xmath77 into the domain wall .
we now have a domain wall excitation @xmath98 as denoted by the fusion rules .
after fusing one of the fluxes from the region with @xmath37 into the domain wall we are left with an isolated flux connected to the domain wall by a string .
if we now create a new pair of fluxes in the region with @xmath37 , it is easy to see that we can move one of these fluxes and fuse it to the vacuum of @xmath37 to end up in a state similar to the one we started with the difference being in the location of the isolated flux connected to the domain wall by a longer string .
this is similar to the presence of a sink for the fluxes .
thus we have synthetically simulated the effect of a sink for the fluxes by creating isolated fluxes connected to the domain walls .
it is clear that such effective sinks can be created in any part of the region with @xmath37 by applying the appropriate operators on the domain wall states . in a similar fashion
we can create sinks for charges using charge excitations instead of flux excitations .
we now show how to create a domain wall state whose effect on the deconfined anyons is to annihilate them .
we start with a ground state , @xmath100 .
we now create a domain wall state by acting with the following operator on a region @xmath102 : @xmath116 now create a pair of fluxes in the region outside @xmath102 where @xmath37 using the following string operator : @xmath117 the domain wall created by @xmath118 can be seen to act as an annihilator when the string @xmath63 crosses @xmath103 as can be seen from the identity @xmath119 this implies that there is a domain wall that can also annihilate the fluxes in region 1 . in the language of states
this can be written as @xmath120 when the dual string crosses the domain wall .
we now show that domain wall created by the operator @xmath118 defined above acts as a scatterer for the fluxes created in the region with @xmath38 states by changing them to the fluxes in the region with @xmath37 . for this
we create two fluxes with a string operator in the region with @xmath38 .
to do this we use the following operator defined along a dual string @xmath121 using the identity @xmath122 we see that when the string in the region @xmath38 crosses @xmath103 it gets flipped to a flux in the @xmath37 region . after it crosses the domain wall
we make it a string which creates @xmath77 , that is @xmath123 .
the resulting state is the one shown in figure [ cond - c ] .
create a domain region by applying the following operator @xmath124_i.\ ] ] compared to the earlier cases acting @xmath106 on @xmath100 results in a different state . in some sense we can think that this dualizes the lattice in region @xmath102 .
so the @xmath32 operators in region @xmath102 now become @xmath125 operators .
we show the explicit action on @xmath100 , @xmath126 with @xmath127 .
if we now create a pair of fluxes with a string operator in a region with @xmath37 that also crosses into the region with states in @xmath38 , created by the operator @xmath106 , we see that this string gets flipped into a string creating charges in the @xmath38 region .
that is the operator which excites fluxes in sector 1 will excite charges in sector 2 after crossing the domain wall .
thus we can create domain walls and domain regions which also switch between anyon types of the two topologically ordered phases .
the local degrees of freedom of the model we have just presented are the elements of @xmath128 .
this space can also be written as a direct sum @xmath129 , which means this model can be thought of as a direct sum of two toric codes and , in this case , the two phases ( @xmath37 and @xmath38 ) do not represent two different topological phases since they have the same set of anyons satisfying the same fusion rules and braiding statistics .
we can generalize the model for the direct sum of two qdms of two groups @xmath1 and @xmath2 , by defining the local hilbert space as @xmath130 , whose basis is @xmath131 .
the hamiltonian is the same as , with the plaquettes and vertex operator as @xmath44 and @xmath132 .
but now the operators @xmath133 and @xmath134 acts as the plaquette and vertex operator of the qdm of the group @xmath135 if all the local state in @xmath89 and in @xmath136 has @xmath37 , otherwise they project the state to the zero vector .
just as in the @xmath137 case , there are two kinds of ground states which we represented by @xmath138 where @xmath139 is the identity of the group @xmath140 . of course there are more states that can be obtained by each of the states above and it is not hard to convince ourselves that this model has one ground state for each ground state of qdm of @xmath1 and another one for each ground state of @xmath2 , and therefore its ground state degeneracy is given by @xmath141 , where @xmath142 is the ground state degeneracy of the qdm of @xmath135 . from the ground state
@xmath68 one can create all the anyonic excitations of the qdm of @xmath140 by applying the ribbon operators ( see @xcite for details ) , and thus all the anyonic excitations of the these two qdms can be realized in this model . in the @xmath129 case
we were able to create zero flux and zero charges excitations , corresponding to the states with mismatch of diamond colors in @xmath89 and/or @xmath136 , here we can create zero flux and charges as well .
we just have to apply the operator @xmath92 that flips the diamond color on a link in order to create these kind of excitations , or one could apply this operator in all the links which are inside an area @xmath102 in order to create a domain wall , as the one shown in figure [ domain - c ] .
it is a known fact @xcite that the anyonic excitations of the qdms of @xmath5 ( and so its topological phases ) are given by irreducible representations of the quantum double algebra @xmath143 .
so if one create a domain wall as the shown in figure [ domain - c ] the excitations which are in the phase @xmath7 are given by the irreducible representations of @xmath144 , but they all can be condensed into the domain wall with the same fusion rules shown in equations and . finally , the two topological phases , corresponding to topological phase of qdm(@xmath1 ) and qdm(@xmath2 ) can be simultaneously realized in the same lattice as long as there is a domain wall separating these two phases , and since the groups @xmath1 and @xmath2 do not need to be the same , the two coexisting phases can be different topological phases . as in the earlier case
we can annihilate anyons at the domain wall and also switch between anyon types by creating appropriate domain walls .
the model we have presented can realize the topological phases of two different qdm of the kitaev type .
moreover , this model can realize these two topological phases either separately or coexisting as long as there is a domain wall formed of zero fluxes and zero charges separating them .
these domain walls are not like the usual ones , as for example the ones where the two dimensional surface where the system lives has a physical boundary @xcite .
they are part of the spectrum of the theory and thus can be created , extended and even destroyed ( with an energy cost proportional to the path length ) , although the excitations living on the domain wall can not be moved freely and can not be fused with another excitation living also in the domain wall .
nevertheless they can be fused with the anyons excitations , as shown in equations and . in other words
, the anyon excitations can get condensed when they are moved into the domain wall as show in figure [ cond - b ] .
therefore , the domain wall is what allows the two different phases to coexist .
this model can be thought of as a direct sum of two qdms , it means the degrees of freedom are either elements of the group @xmath1 or @xmath2 , and they are controlled by the addition of an extra quantum number @xmath80 .
this construction may seem artificial at first glance , but there is a strong algebraic structure behind it , namely groupoid algebras . just as the group algebra @xmath145 of a given group @xmath5 is the main algebraic structure behind the qdm of the kitaev type ,
the main structure behind this model is the groupoid algebra @xmath146 , which is very similar to the group algebra apart from the fact that @xmath147 is a groupoid instead of a group .
nevertheless the groupoids we are considering are of a special kind , it is formed by two disconnected groups as follows : let @xmath148 .
the product of two elements @xmath149 is defined if , and only if , @xmath150 , and in this case the product is the usual group product @xmath151 .
the groupoid algebra @xmath146 , whose basis is @xmath152 is defined as @xmath153 the local hilbert space @xmath130 can be defined as @xmath154 . unlike the group algebra ,
the groupoid algebra is not a hopf algebra , it is an example of a weak hopf algebra @xcite .
note that the groupoid algebras we have considered are not similar to the direct product of two groups .
the zeros in the groupoid algebra is the main difference between the two cases as it does not exist in the definition of the direct product of two groups .
thus if we construct a kitaev type qdm based on the direct product of two groups we will not see the domain walls that we saw in the qdm based on the groupoid algebra . in @xcite we have showed how to construct the quantum double of a given input using the construction of kuperberg s invariant @xcite .
we can think of the model in this paper as choosing that input to be that of a groupoid algebra .
a more general construction can be done if one consider @xmath147 as being any groupoid .
we will report features of these models in an accompanying paper .
some of these features include generating domain walls between a topologically ordered phase containing deconfined excitations and a phase where these excitations are confined with string tension terms .
models with string tension have been considered earlier in @xcite .
quantum double models based on a particular type of quantum groupoids have been considered earlier in @xcite .
defects in topologically ordered phases have gained wide attention @xcite .
these papers describe the possible topological order that emerges when one considers defects ( or an external symmetry @xmath5 ) in the system by studying in detail the theory of @xmath5 crossed braided extensions of the parent fusion category .
however there is still a dearth of exactly solvable models that describe these topologically ordered phases with defects .
some early works include the example provided by bombin @xcite and simple generalizations of the same @xcite .
the theory of defects were studied further in @xcite .
tunneling between topologically ordered phases were also studied in @xcite .
it will be interesting to see if the models constructed here describe similar phases to the ones studied earlier .
x. chen , z.- c. gu , z.- x. liu , x.- g. wen , _ symmetry protected topological orders and the group cohomology of their symmetry group _ , phys . rev .
b 87 , 155114 ( 2013 ) and arxiv:1106.4772 [ cond-mat.str-el ] .
j. eom , h. cho , w. kang , k. l. campman , a. c. gossard , m. bichler , w. wegscheider , _ quantum hall ferromagnetism in a two - dimensional electron system _ , science 289 , 2320 ( 2000 ) and arxiv : cond - mat/0011081 [ cond-mat.mes-hall ] .
p. padmanabhan , j. p. i. jimenez , m. j. b. ferreira , p. teotonio - sobrinho , _ a recipe for constructing frustration - free hamiltonians with gauge and matter fields in one and two dimensions _ , to appear in jphysa and arxiv:1503.07601 [ cond-mat.str-el ] . | transitions between different topologically ordered phases have been studied by artificially creating boundaries between these gapped phases and thus studying their effects relating to condensation and tunneling of particles from one phase to the other . in this work
we introduce exactly solvable models which are similar to the quantum double models ( qdm ) of kitaev where such domain walls are dynamically generated making them a part of the spectrum .
these systems have a local symmetry and may or may not have a global symmetry leading to the possibility of a ground state degeneracy arising from both , global symmetry breaking and a topological degeneracy .
the domain wall states now separate the different topologically ordered phases belonging to the different sectors controlled by the global symmetry , when present .
they have interesting properties including fusion with the deconfined anyons of the topologically ordered phase , to either create new domain walls , or their annihilation
. they can also act like a scatterer permuting anyons of the topologically ordered phases on either side of the domain wall .
thus these domain wall states can be thought of as a synthetic scatterer which can be created in any part of the lattice .
we show these effects for the simplest case of the @xmath0 toric code phase decorated with a global @xmath0 symmetry and then make remarks about the case when we have the qdm based on two arbitrary groups @xmath1 and @xmath2 in which case we may no longer have a global symmetry .
@xmath3 departamento de fsica matemtica , universidade de so paulo rua do mato travessa r nr.187 , so paulo , cep 05508 - 090 @xmath4 department of physics and astronomy , university of california , irvine 4129 frederick reines hall irvine , ca 92697 - 4575 |
dissipative hydrodynamics has been applied to relativistic heavy ion collisions to understand the viscous nature of the quark gluon plasma ( qgp ) , in particular , to explain the experimental data of the collective flow parameter @xmath4 by including the effect of shear viscosity @xcite .
one of the objectives of these attempts is to determine @xmath5 , where @xmath6 is the shear viscosity coefficient and @xmath7 is the entropy density , by comparing hydrodynamical calculations of @xmath8 with the measured values .
however , in addition to the uncertainties on the initial condition for hydrodynamics and the breakdown of the traditional cooper - frye freeze - out procedure in dissipative hydrodynamics @xcite , there are several difficulties in these attempts .
one should note that these studies are mainly based on three assumptions ; 1 ) the effects of other irreversible currents , such as the bulk viscous pressure and heat conductivity , are smaller than those of the shear stress tensor , 2 ) @xmath9 does not strongly depend on temperature and 3 ) the effects of shear viscosity on observables are mostly characterized by one transport coefficient , @xmath5 . in this work , we would like to discuss the validity of these assumptions .
first , @xmath5 is in general a function of temperature . while the exact temperature dependence of this ratio is still unknown ,
its qualitative behavior is somehow well understood . as was shown in ref .
@xcite , @xmath10 exhibits a complex temperature dependence showing a minimum near the qcd phase transition .
thus , it is not clear that such a temperature dependence can be approximately replaced by a constant value .
second , we remark that the effects of the shear stress tensor ( or any other irreversible current ) can not be characterized only by @xmath0
. this might be the case for a fluid described by the relativistic navier - stokes ( rns ) equation .
however , as was shown in refs .
@xcite , hydrodynamical equations which violate causality , such as the rns equations , are unstable and can not be used for any application . to solve the problem of acausality and instability
, we have to introduce ( at least ) one more transport coefficient called the relaxation time @xmath11 , which represents the retardation effect in the formation of the irreversible current . then it is natural to expect that the relaxation time should affect the behavior of @xmath12 or any other observable .
because of the relaxation time , a large @xmath13 does not necessarily induce a large shear stress tensor . to illustrate this ,
let us consider the retardation effect of an arbitrary irreversible current @xmath14 which is induced by a thermodynamic force @xmath15 .
for the sake of simplicity , we consider a constant @xmath15 .
then , the current is given by @xmath16where @xmath17 and @xmath11 are the transport coefficients for this respective irreversible current .
one can see that if @xmath11 is small enough compared to the typical scale of the hydrodynamic evolution , @xmath14 is simply proportional to @xmath18 .
however , if @xmath11 is not small , we have to consider the competition of @xmath17 and @xmath11 to quantify the dissipative effects . as a matter of fact , the time scale of the hydrodynamic evolution in relativistic heavy - ion collisions is of the order of 10 fm . on the other hand , as is shown in ref .
@xcite , the relaxation time of the shear viscosity @xmath19 can also reach values around 10 fm . in this sense
, we should not ignore the effect of @xmath19 in the analysis of @xmath8 . in this paper
, we calculate the effect of the shear and bulk viscosities on elliptic flow by taking a realistic parameterization of the shear and bulk viscous coefficients , @xmath0 and @xmath1 , and the corresponding relaxation times , @xmath2 and @xmath3 .
we argue that the behaviors close to ideal fluid observed at rhic energies may be related to non - trivial temperature dependence of these transport coefficients .
in this work , we will use the memory function method @xcite , which is one of the formulations of relativistic dissipative hydrodynamics . for a general metric
, the conservation of energy and momentum is expressed as @xmath20 where @xmath21 denotes the covariant derivative .
we use the landau definition of the four - fluid velocity @xmath22 in which case the energy - momentum tensor is decomposed as @xmath23 , where , @xmath24 , @xmath25 , @xmath26 and @xmath27 are the energy density , pressure , bulk viscous pressure and shear stress tensor , respectively . the projection operator is defined as usual , @xmath28 . in the memory function method ,
any irreversible current @xmath14 is induced by a thermodynamic force @xmath15 as @xmath29this equation is derived to take into account time retardation effects and the extensivity of hydrodynamic variables .
one can regard this equation as the generalized version of the maxwell - cattaneo equation .
it should be emphasized that the existence of the non linear term @xmath30 is essential to obtain a stable theory as is discussed in ref .
@xcite . by applying this equation to the cases of the shear stress tensor and the bulk viscous pressure , we obtain @xmath31where @xmath32 and @xmath33 are the thermodynamic forces .
we use the traditional notation @xmath34 where @xmath35 is the symmetric traceless double projection operator .
the above hydrodynamical equation are solved numerically in three spatial dimensions .
we use the smoothed particle hydrodynamics ( sph ) method .
@xcite for details .
we use approximately 80000 sph particles with a smoothing parameter @xmath36 fm @xcite . for simplicity , we use a factorized initial energy density profile into its longitudinal and transverse parts , as proposed by ref .
the transverse part is parameterized as in ref .
the longitudinal part is taken as in ref .
@xcite , with the choice of parameters @xmath37 and @xmath38 .
the initial value of the irreversible currents will be taken as zero and the impact parameter is always set as @xmath39 fm .
we use the equation of state ( eos ) introduced in @xcite . as for the freeze out
, we used the ordinary cooper - frye formula @xcite .
however , we did not consider non - equilibrium corrections to the one - particle distribution function , because , as was pointed out in refs .
@xcite , this correction gives rise to unphysical behaviors and is not reliable .
however , this will not affect the conclusions of this paper . as mentioned in the introduction ,
the temperature dependence of the transport coefficients is very important .
the microscopic formula to calculate these quantities is given in ref .
however , the application of this formula is still limited to leading order perturbative calculations .
fortunately , in this approximation @xmath13 and @xmath40 become equivalent to the ones calculated in the green - kubo - nakano formalism and we can use these results . in this paper , we use the lattice qcd calculations @xcite for the qgp phase and we use the results from ref .
noronha for the hadron phase .
these calculations for @xmath5 were fitted and the result is shown in fig .
[ shear_coef](a ) .
we can see that @xmath9 shows a minimum around @xmath41 mev . for the bulk viscosity coefficient @xmath42
, we use the parameterization showed in ref .
@xcite with a critical temperature of @xmath41 mev .
( left panel ) and the shear relaxation time @xmath43 ( right panel).,title="fig : " ] [ shear_coef ] ( left panel ) and the shear relaxation time @xmath43 ( right panel).,title="fig : " ] [ shear_tau ] similarly , in leading order calculations , the microscopic formula predicts a simple relation for the shear viscosity , @xmath44 . for the bulk viscosity , there is still no result and we simply assume @xmath45 . since @xmath19 ( @xmath46 ) is linear with the transport coefficient @xmath13 ( @xmath40 ) , the critical behavior of @xmath6 ( @xmath40 ) is transfered to that of @xmath19 ( @xmath46 ) .
thus , @xmath19 will have a minimum around the qcd phase transition while @xmath46 will display a maximum . in this sense ,
the complex temperature dependences of the relaxation times are induced by those of @xmath47 and @xmath1 .
calculated for different choices of @xmath48 ( left panel ) and calculated including the effects of bulk viscosity ( right panel ) .
, title="fig : " ] [ effective_vis ] calculated for different choices of @xmath48 ( left panel ) and calculated including the effects of bulk viscosity ( right panel ) .
, title="fig : " ] [ bulk ] in fig .
[ effective_vis](a ) , we show @xmath49 for different choices of @xmath9 and neglecting bulk viscosity .
these calculations were made for thermal pions with a freeze - out temperature of @xmath50 mev .
the solid line corresponds to the case where @xmath5 has the temperature dependence shown in fig .
[ shear_coef](a ) whereas the dashed , dashed - dotted and dotted lines correspond to constant values of @xmath5 of 0.08 , 0.12 and 0.16 , respectively .
one can see that the elliptic flow calculated with the temperature dependent @xmath5 can be well reproduced by the one calculated with an effective constant shear viscosity coefficient @xmath51 .
the quantity @xmath52 is very close to the minimum of @xmath9 .
this can be explained by the behavior of the relaxation time , shown in fig .
shear_tau(b ) . as mentioned in the introduction
, @xmath19 characterizes the time scale for the velocity gradients to be converted to viscosity . because the fluid expands with comparable time scale with @xmath53 , the magnitude of the shear stress tensor
does not become large despite the large values of @xmath9 achieved throughout the hydrodynamical evolution .
it should be noted that @xmath52 is not bounded by the minimum of @xmath54 .
let us consider another set of parameters , @xmath55 . in this case , the effective viscosity @xmath56 becomes smaller than the minimum value of @xmath5 .
this result shows that the value of @xmath57 is not a direct measurement of @xmath5 .
it can come from the combined effect of @xmath5 and @xmath2 , and the small values of @xmath57 found at rhic can just as well be a manifestation of the large relaxation time in the hadronic and qgp phases . whether @xmath52 can be used to quantify the value of viscosity at the phase transition or not , depends on the magnitude of the relaxation time in these regions . in fig .
[ bulk](b ) , we show @xmath49 considering the combined effects of bulk and shear viscosities , with realistic transport coefficients .
the solid line corresponds to the calculation with only shear viscosity , the dashed line to the calculation with only bulk viscosity and the dashed - dotted line to the calculation with both bulk and shear viscosities .
we see that the effect of bulk viscosity is rather small despite the large values of @xmath58 near the phase transition .
this can be explained as an effect of the relaxation time @xmath46 , again .
since @xmath59 also becomes large there , the values of @xmath60 do not generate a large bulk viscous pressure .
since @xmath40 decreases very quickly outside the phase transition region , we can not see considerable effects of the bulk viscosity .
we remark that all the calculations showed in this work were done in three spatial dimensions .
however , we were mainly interested in @xmath8 at zero rapidity , where we confirmed that the bjorken scaling @xmath61 is a good approximation .
thus , all the results showed here are approximately equivalent to ( 2 + 1)d hydrodynamical calculations with the bjorken scaling for the longitudinal direction .
this happened mainly because of our choice of initial condition , which displays a plauteu in energy density in the longitudinal direction .
studies with different initial conditions , specially in the longitudinal direction , will be done in future .
we have shown that due to the interplay between the effects of viscosity and relaxation time , it is possible to fit the behavior of collective flow in terms of one unique effective shear viscosity @xmath56 .
however , its value depends critically on the evolution of the system .
specifically , the value is affected by the time span for which the system remains in different phases . for the lhc energies , for example
, we would expect a different value for the effective viscosity since the system will stay longer in the quak gluon plasma phase .
however , the details of such changes are hard to calculate due to the uncertainties on @xmath13 .
we also remark that the uncertainties on the relaxation time are even larger since not so many effort have been applied to calculate the temperature dependence of this transport coefficient ( unlike the @xmath5 case ) .
we acknowledge discussions with j. noronha , p. huovinen , r. lacey and d. h. rischke . this work has been supported by cnpq , faperj , capes , pronex and the helmholtz international center for fair within the framework of the loewe program ( landesoffensive zur entwicklung wissenschaftlich- okonomischer exzellenz ) launched by the state of hesse . | in this work , we examine the effect of shear and bulk viscosities on elliptic flow by taking a realistic parameterization of the shear and bulk viscous coefficients , @xmath0 and @xmath1 , and their respective relaxation times , @xmath2 and @xmath3 .
we argue that the behaviors close to ideal fluid observed at rhic energies may be related to non - trivial temperature dependence of these transport coefficients . |
the skyrme model @xcite has been discussed for more than 40 years .
it is an effective field theory describing hadrons @xcite .
it is defined by the lagrangian density @xmath1+\frac{c_4}{2}\mbox{tr}\left([g^{\dagger}\partial_{\mu}g\,,\,g^{\dagger}\partial_{\nu}g][g^{\dagger}\partial^{\mu}g\,,\,g^{\dagger}\partial^{\nu}g]\right ) , \label{eqn : lagrangian}\end{aligned}\ ] ] where @xmath2 is an element of @xmath3(2 ) and @xmath4 and @xmath5 are coupling constants .
if we define @xmath6 and @xmath7 by @xmath8 @xmath9 is expressed as @xmath10 where @xmath11 are pauli matrices .
the field equation is given as the conservation law @xmath12 where @xmath13 @xmath14 are defined by @xmath15 the field @xmath16 is proportional to the isospin current of the model .
another important conserved current is the baryon number current @xcite : @xmath17 the conservation law @xmath18 follows solely from the definition of @xmath19 irrespective of the field equation for @xmath6 . by definition
, @xmath6 satisfies the condition @xmath20 it was discussed @xcite that the skyrme model is intimately related to the faddeev model @xcite defined by @xmath21 where @xmath22 is a three - component scalar field satisfying @xmath23 and @xmath24 is given by @xmath25 the field equation for @xmath26 is given by @xmath27 the faddeev model is expected to describe the low energy gluonic dynamics of qcd @xcite . if we define @xmath28 and @xmath29 by @xmath30 with @xmath31 eqs .
( [ eqn : conservationlow ] ) , ( [ eqn : conditiona ] ) , and ( [ eqn : equationn ] ) become @xcite @xmath32=0 , \label{eqn : motioneq}\\ & \partial_{\mu}{\mbox{\boldmath $ a$}}_{\nu}-\partial_{\nu}{\mbox{\boldmath $ a$}}_{\mu}=2{\mbox{\boldmath $ a$}}_{\mu}\times{\mbox{\boldmath $ a$}}_{\nu } , \\ & \partial_{\mu}\left[c_2{\mbox{\boldmath $ b$}}^{\mu}+c_4\left({\mbox{\boldmath $ b$}}^{\mu}\times{\mbox{\boldmath $ b$}}^{\nu}\right)\times{\mbox{\boldmath $ b$}}_{\nu}\right]=0.\end{aligned}\ ] ] from the definition ( [ eqn : defbb ] ) , we have the condition @xmath33 for @xmath34 .
the parallelism of @xmath34 with @xmath35 is evident .
of course , the two models are different from each other because the degree of freedom of the skyrme model is 3 while that of the faddeev model is 2 .
it can be seen , however , that the configuration @xmath34 yields a restricted class of the configuration @xmath35 .
the soliton solutions of the skyrme model are classified by the baryon number @xmath36 and are identified with baryons @xcite . on the other hand ,
the soliton solutions of the faddeev model are classified by the topological number @xmath37 called the hopf charge and are identified with glueballs @xcite .
the numerical analysis of these models revealed rich spectra of solitons : knot solitons for the faddeev model and polyhedral solitons for the skyrme model @xcite .
for example , the solution of the faddeev model with @xmath38 is the soliton of the trefoil knot structure . as for the analytic solutions for these models ,
only a few simple examples are known .
skyrme @xcite found that the configuration @xmath39 with @xmath40 leading to the field @xmath6 of the form @xmath41 satisfies the field equation if @xmath42 is light - like : @xmath43 . in a recent paper @xcite , two of the present authors ( m.h . and j.y . )
obtained a solution of the form @xmath44 with @xmath45 .
although the solution ( [ eqn : kf ] ) is independent of the coupling constants @xmath4 and @xmath5 in @xmath9 , the solution ( [ eqn : kl ] ) depends on them through the dimensionless combination @xmath46 in this paper , we seek a solution of the form @xmath47 with @xmath48 it turns out that the solution can be described with the help of the function @xmath49 where @xmath50 is the weierstrass @xmath0 function satisfying the differential equation @xmath51 ^ 2=4\left[\wp(z)-e_1\right]\left[\wp(z)-e_2\right]\left[\wp(z)-e_3\right ] .
\label{eqn : wpfunc}\end{aligned}\ ] ] here @xmath52 and @xmath53 are real constants satisfying @xmath54 , sn@xmath55 is the jacobi elliptic function of @xmath56 with the modulus @xmath57 , @xmath58 is the second fundamental period of @xmath50 , and @xmath59 is a linear combination of @xmath60 and @xmath61 .
we note that @xmath62 is equal to @xmath63 multiplied by a constant independent of the momenta @xmath57 , @xmath64 and @xmath65 .
in contrast with the solutions ( [ eqn : kf ] ) and ( [ eqn : kl ] ) with the vanishing baryon number density , our solution ( [ eqn : klm ] ) possesses the nonvanishing baryon number density @xmath66 where @xmath67 is a constant independent of @xmath68 and @xmath65 .
this paper is organized as follows . in sec.[sec : formulation of the problem ] , we seek a convenient set of variables and fields and write the field equation in a compact way . in sec.iii , we introduce a set of _ anstze _ for @xmath35 compatible with the field equation . by the above _ anstze _ ,
the non linear partial differential equations for @xmath35 are reduced to some algebraic constraints for parameters specifying solutions .
some examples of the solutions of the above algebraic constraints are given .
section v is devoted to a summary and discussion .
in order to discuss the case of @xmath69 , we introduce the parameters @xmath70 where the triplet @xmath71 should be ( 1,2,3 ) or ( 2,3,1 ) or ( 3,1,2 ) and @xmath72 , @xmath73 , @xmath74 .
then we have @xmath75 we now introduce @xmath76 and write @xmath77 , @xmath78 and @xmath79 . for @xmath80 , @xmath81 in eq .
( [ eqn : amug ] ) can be written as @xmath82 if we define @xmath83 , @xmath84 and @xmath85 by @xmath86 @xmath28 can be rewritten as @xmath87 we find that the integrability condition ( [ eqn : conditiona ] ) becomes @xmath88 which yields @xmath89 the second term on the right hand side of eq .
( [ eqn : jmu ] ) is expressed as @xmath90 , \label{eqn : lhs}\end{aligned}\ ] ] where @xmath91 , @xmath92 , @xmath93 and @xmath94 are given by @xmath95 therefore , the field equation ( [ eqn : conservationlow ] ) becomes @xmath96 it is a vector nonlinear differential equation .
it is difficult to discuss it generally , so we will introduce some _ anstze _ in the next section .
in order to obtain the solutions of eqs .
( [ eqn : condition ] ) and ( [ eqn : eqmotion1 ] ) , we first assume @xmath97 with @xmath98 and @xmath99 . here
@xmath100 are constants .
since the condition ( [ eqn : condition ] ) can be written as @xmath101 we are led to the parametrization @xmath102 it is interesting to note that the three conditions @xmath103 for @xmath104 and @xmath105 are satisfied if @xmath106 and @xmath107 satisfy the single relation @xmath108 in addition to eq .
( [ eqn : ansatz1 ] ) , we assume @xmath109 where @xmath110 and @xmath111 are constants . then , the conditions ( [ eqn : dcondition ] ) yield @xmath112 it can be seen that the eight conditions ( [ eqn : constraints1 ] ) and ( [ eqn : constraints7 ] ) for nine parameters are equivalent to five conditions @xmath113 we now regard the independent parameters as @xmath114 and @xmath115 .
the other parameters @xmath116 and @xmath117 are given as @xmath118 our _ anstze _ proposed in the previous subsection readily lead us to the consequence that @xmath119 and @xmath120 are constant .
we also find the relations @xmath121 from the consistency conditions @xmath122 we obtain @xmath123}{\partial\eta}+\sigma\frac{\partial\left[{\mbox{\boldmath $ a$}}\cdot({\mbox{\boldmath $ b$}}\times{\mbox{\boldmath $ c$}})\right]}{\partial\xi}&=\rho\frac{\partial\left[{\mbox{\boldmath $ a$}}\cdot({\mbox{\boldmath $ b$}}\times{\mbox{\boldmath $ c$}})\right]}{\partial\zeta}+\mu\frac{\partial\left[{\mbox{\boldmath $ a$}}\cdot({\mbox{\boldmath $ b$}}\times{\mbox{\boldmath $ c$}})\right]}{\partial\eta } \nonumber\\ & = \nu\frac{\partial\left[{\mbox{\boldmath $ a$}}\cdot({\mbox{\boldmath $ b$}}\times{\mbox{\boldmath $ c$}})\right]}{\partial\xi}+\lambda\frac{\partial\left[{\mbox{\boldmath $ a$}}\cdot({\mbox{\boldmath $ b$}}\times{\mbox{\boldmath $ c$}})\right]}{\partial\zeta } \nonumber \\ & = 0.\end{aligned}\ ] ] with the aid of the relations ( [ eqn : relation5 ] ) and ( [ eqn : adotbc ] ) , we conclude that @xmath124 and @xmath125 are given in the following way : @xmath126 where @xmath127 is a function of @xmath128 to be determined and @xmath129 , @xmath130 and @xmath131 are arbitrary constants .
the function @xmath127 is fixed as follows . from the formula @xmath132 ^
2=\left|\begin{array}{ccc}{\mbox{\boldmath $ a$}}^2 & { \mbox{\boldmath $ a$}}\cdot{\mbox{\boldmath $ b$}}&{\mbox{\boldmath $ a$}}\cdot{\mbox{\boldmath $ c$}}\\ { \mbox{\boldmath $ a$}}\cdot{\mbox{\boldmath $ b$}}&{\mbox{\boldmath $ b$}}^2 & { \mbox{\boldmath $ b$}}\cdot{\mbox{\boldmath $ c$}}\\ { \mbox{\boldmath $ a$}}\cdot{\mbox{\boldmath $ c$}}&{\mbox{\boldmath $ b$}}\cdot{\mbox{\boldmath $ c$}}&{\mbox{\boldmath $ c$}}^2 \end{array}\right|\end{aligned}\ ] ] and eqs .
( [ eqn : ab ] ) and ( [ eqn : p ] ) , we observe that @xmath133 defined by @xmath134 should satisfy an equation of the form @xmath135
^ 2&=4\left[k(\omega)\right]^3-g_2k(\omega)-g_3 \\ & = 4\left[k(\omega)-e_1\right]\left[k(\omega)-e_2\right]\left[k(\omega)-e_3\right],\nonumber\end{aligned}\ ] ] where the constants @xmath136 and @xmath137 are complicated functions of @xmath138 and @xmath131 and satisfy the relations @xmath139 this equation indicates that @xmath133 is given by the weierstrass @xmath0 function as @xmath140 .
we hereafter assume that the constants @xmath138 and @xmath131 are chosen so that the inequality @xmath141 holds .
then the constants @xmath52 and @xmath53 are real and can be assumed to satisfy @xmath54 . to remove the poles for the real values of @xmath128
, we adopt @xmath133 given by eq .
( [ eqn : komega ] ) . the explicit formula for @xmath142
is given by @xmath143 we find that @xmath142 is a purely imaginary number .
other choices of the integral constant lead to a complex - valued @xmath133 . from eqs .
( [ eqn : xi ] ) and ( [ eqn : omega ] ) , @xmath128 is calculated as @xmath144 .
\label{eqn : l}\end{aligned}\ ] ] as was noted below eq .
( [ eqn : wpfunc ] ) , @xmath62 is a constant multiple of @xmath63 : @xmath145\frac{\mu}{\nu^3}\frac{c_2}{c_4}.\end{aligned}\ ] ] if we define the differential operator @xmath146 by @xmath147 @xmath148 and @xmath128 satisfy @xmath149 since there exists another independent variable @xmath150 satisfying @xmath151 , e.g. @xmath152 @xmath153 and @xmath85 are represented as @xmath154 we have thus seen that the general solution of our _ ansatz _ are given by eqs .
( [ eqn : l ] ) , ( [ eqn : omega ] ) and ( [ eqn : babbbc ] ) .
we note that we have not yet made use of the field equation ( [ eqn : eqmotion1 ] ) . in the next section
, we shall see that the field equation can be used to constrain the allowed values of the independent parameters @xmath114 and @xmath115 .
the baryon number current @xmath19 in the skyrme model was defined by eq.([eqn : baryonn ] ) .
substituting eq .
( [ eqn : amu ] ) into eq .
( [ eqn : baryonn ] ) , we obtain the baryon number density ( [ eqn : baryonnform ] ) with @xmath67 given by @xmath155
to discuss the field equation , it is convenient to introduce @xmath156 and @xmath157 by @xmath158 then the field equation ( [ eqn : eqmotion1 ] ) becomes @xmath159 with the help of the _ anstze _ ( [ eqn : ansatz1 ] ) and ( [ eqn : ansatz2 ] ) , @xmath156 and @xmath157 can be expressed solely by @xmath83 , @xmath84 and @xmath85 without their derivatives .
after some manipulations , we obtain @xmath160\left(a{\mbox{\boldmath $ a$}}+b{\mbox{\boldmath $ b$}}+c{\mbox{\boldmath $ c$}}\right)+d({\mbox{\boldmath $ a$}}\times{\mbox{\boldmath $ b$}})+e({\mbox{\boldmath $ b$}}\times{\mbox{\boldmath $ c$}})+f({\mbox{\boldmath $ c$}}\times{\mbox{\boldmath $ a$}}),\end{aligned}\ ] ] where @xmath161 , @xmath162 , @xmath163 , @xmath146 , @xmath164 and @xmath165 are given by @xmath166 assuming that @xmath125 is nonvanishing , eq .
( [ eqn : rq ] ) is equivalent to the condition @xmath167 where @xmath168 , @xmath169 , @xmath170 are defined by @xmath171 by making use of eq .
( [ eqn : ab ] ) , we see that @xmath168 , @xmath169 and @xmath170 take the following form : @xmath172 where @xmath173 , @xmath174 , @xmath56 , @xmath175 , @xmath176 , @xmath177 are constant .
we now find that the field equation has been reduced to the equations @xmath178 we first discuss the equations @xmath181 .
the constants @xmath175 , @xmath176 and @xmath177 in eq .
( [ eqn : mastu ] ) are expressed as @xmath182 where @xmath183 , @xmath184 and @xmath185 are given by @xmath186 to make @xmath175 , @xmath176 and @xmath177 vanishing , we have only to choose the arbitrary constants @xmath129 , @xmath130 and @xmath131 as @xmath187 we next consider the three equations @xmath189 . it can be seen that they are equivalent to the following algebraic equations for @xmath111 and @xmath115 : @xmath190 where @xmath191 @xmath192 , @xmath193 , @xmath194 , @xmath195 and @xmath196 are given by @xmath94 and @xmath197 as follows : @xmath198 , \\
f_1&\hspace{-2mm}=-\beta(2+\beta)(2+\alpha+\beta ) , \end{array } \\ & \begin{array}{ll } p_2&\hspace{-2mm}=\alpha(2+\alpha ) , \\
q_2&\hspace{-2mm}=(2+\alpha+2\beta)(4+\alpha+2\beta ) , \\ r_2&\hspace{-2mm}=\alpha(2+\alpha)(2+\alpha+\beta)^2 , \\ h_2&\hspace{-2mm}=(2+\alpha)^2 + 2(1+\alpha)\beta , \\
g_2&\hspace{-2mm}=\alpha(2+\alpha)(2+\alpha+\beta ) , \\ f_2&\hspace{-2mm}=-(2+\alpha+\beta)\left[(2+\alpha)^2 + 2(1+\alpha)\beta\right ] , \end{array } \\ & \begin{array}{ll } p_3&\hspace{-2mm}=1 , \\
q_3&\hspace{-2mm}=1 , \\ r_3&\hspace{-2mm}=-4+(\alpha-\beta)^2 , \\
h_3&\hspace{-2mm}=-1 , \\ g_3&\hspace{-2mm}=\beta-\alpha , \\ f_3&\hspace{-2mm}=\beta-\alpha . \end{array}\end{aligned}\ ] ] the parameters @xmath94 and @xmath197 must be chosen so that the simultaneous equations ( [ eqn : fi ] ) of second order for @xmath111 and @xmath115 possess a common root @xmath199 . to obtain the condition that @xmath94 and @xmath197 must satisfy , assuming @xmath200 , @xmath201 , @xmath202 , we rewrite the equation @xmath203 as follows : @xmath204 where @xmath205 , @xmath206 , @xmath207 and @xmath208 @xmath209 are defined by @xmath210 if we eliminate @xmath111 from eqs .
( [ eqn : q ] ) and ( [ eqn : q ] ) , we obtain @xmath211 with @xmath212 if we eliminate @xmath111 from eqs .
( [ eqn : q ] ) and ( [ eqn : q ] ) , we obtain @xmath213 with @xmath214 the condition for the two equations ( [ eqn : a ] ) and ( [ eqn : b ] ) to have a common root @xmath115 is that the resultant for these equations vanishes : @xmath215 we note that the condition ( [ eqn : resultant ] ) is necessary but not sufficient for the original equations ( [ eqn : fi ] ) to have a common root @xmath199 . in general
, the condition ( [ eqn : resultant ] ) is a very complicated relation between @xmath94 and @xmath197 .
it is clear , however , that there exist infinite number of pairs @xmath216 that satisfy ( [ eqn : resultant ] ) . for simplicity ,
let us consider the case @xmath217 .
then eq . ( [ eqn : resultant ] ) becomes @xmath218 where @xmath219 and @xmath220 are defined by @xmath221 we here exclude the case @xmath222 , in which our formulation above should be modified .
we find two real roots @xmath223 of @xmath224 and two real roots @xmath225 of @xmath226 .
we see , however , that only the roots @xmath227 and @xmath228 give rise to the common root @xmath229 for eq .
( [ eqn : fi ] ) . for both @xmath227 and @xmath228 , the common root @xmath199 is given by ( 1 , -1 ) . as was stated below eq .
( [ eqn : resultant ] ) , this is because eq .
( [ eqn : resultant ] ) is a necessary but not a sufficient condition for eq .
( [ eqn : fi ] ) to possess a common root @xmath199 .
in order to obtain @xmath83 , @xmath84 and @xmath85 , we must keep up the inequalities @xmath233 where @xmath234 , @xmath235 and @xmath236 denote @xmath237 , @xmath238 and @xmath239 , respectively . with the help of the inequality @xmath240 , which is valid in the case noted around eq .
( [ eqn : wpinequality ] ) , we easily obtain the maxima and the minima of @xmath241 , @xmath242 and @xmath243 .
they depend on the signatures of @xmath244 , @xmath245 and @xmath246 . for
the above example @xmath247 , @xmath248 , @xmath249 , we have @xmath250 , @xmath251 and obtain the condition @xmath252 where @xmath253 , @xmath254 , @xmath255 , @xmath256 , @xmath257 and @xmath258 are given by @xmath259 sufficient conditions for ( [ eqn : mm1 ] ) to hold valid are @xmath260 where @xmath261 and @xmath262 are defined by @xmath263 and @xmath264 , satisfying @xmath265 .
it can be numerically checked that , e.g. @xmath266 , @xmath267 , @xmath268 satisfy eq .
( [ eqn : ab ] ) .
thus we realize that there indeed exist solutions that we intended to obtain . as for the case
@xmath269 , @xmath248 , @xmath249 , we have @xmath250 , @xmath270 .
we then find that @xmath271 should be satisfied . in this case
, we obtain the sufficient conditions @xmath272 we find that , e.g. @xmath273 , @xmath274 and @xmath275 are the case . we finally discuss how to obtain @xmath83 , @xmath84 , @xmath85 from given @xmath279 , @xmath280 , @xmath120 , @xmath241 , @xmath242 and @xmath243 . expressing @xmath241 , @xmath242 and @xmath243 as @xmath281
we introduce the unit vectors @xmath282 and @xmath283 as @xmath284 comparing eq .
( [ eqn : bcp ] ) with eq .
( [ eqn : ab ] ) , we can regard @xmath285 and @xmath286 as given by @xmath133 .
the angles @xmath287 and @xmath288 can be chosen so as to satisfy @xmath289 we also introduce the unit vectors @xmath290 , @xmath291 and @xmath292 .
they can be assumed to satisfy @xmath293 then we set @xmath294 it is clear that the above @xmath295 , @xmath296 and @xmath297 satisfy ( [ eqn : ab ] ) .
we find that @xmath125 is calculated to be @xmath298 .
the relationship among @xmath83 , @xmath84 , @xmath85 , @xmath299 , @xmath300 , @xmath301 , @xmath302 and @xmath303 can be depicted as in fig.1 .
we note that we still have freedom in choosing @xmath290 and @xmath291 . ,
@xmath277 , @xmath278 , @xmath304 , @xmath305 , @xmath306 , @xmath307 , @xmath308.,width=226 ]
we have obtained solutions of the skyrme model of the form @xmath309 with the momenta @xmath310 satisfying @xmath311 .
we found that the variables @xmath312 and @xmath313 defined by eq .
( [ eqn : xi ] ) and the fields @xmath153 and @xmath85 defined by eq .
( [ eqn : aia ] ) are convenient to describe the problem .
the existence condition for @xmath2 and the field equation were expressed in compact forms as seen in eqs .
( [ eqn : condition ] ) and ( [ eqn : eqmotion1 ] ) . to solve these equations , we introduced the _
anstze _ ( [ eqn : ansatz1 ] ) and ( [ eqn : ansatz2 ] ) for the derivatives of @xmath153 and @xmath85 with respect to @xmath312 and @xmath313 .
the _ anstze _ contained nine parameters @xmath314 and @xmath117 .
it was observed that the consistency of the _ anstze _ yield five constraints ( [ eqn : constraints5 ] ) for the parameters , leaving @xmath114 and @xmath115 independent . through the _ anstze
_ , we found that @xmath119 and @xmath120 are constants , @xmath315 and @xmath243 are given by the weierstrass @xmath0 function , and @xmath125 is given by the derivative of the @xmath0 function . with the help of the _ anstze _ ,
the field equation was reduced to six algebraic constraints for the parameters @xmath316 and @xmath131 , where @xmath317 and @xmath131 are arbitrary constants contained in @xmath315 and @xmath243 , respectively .
three of the six constraints were used to fix @xmath317 and @xmath131 .
the structure of the remaining three constraints for @xmath114 and @xmath115 was discussed .
it was confirmed numerically that there indeed exist solutions of the field equation . from the solutions @xmath153 and @xmath85
, we can construct @xmath6 by eq .
( [ eqn : amu ] ) .
the matrix field @xmath2 is obtained as @xmath318\frac{dx^{\mu}(\lambda)}{d\lambda}\tau^{\alpha}\right],\end{aligned}\ ] ] where @xmath319 parametrizes a path from @xmath320 to @xmath321 , and @xmath322 is the anti-@xmath323 ordering operator .
the value of @xmath2 is independent of the choice of the path because of eq .
( [ eqn : amu ] ) .
the above expression for @xmath2 is rather symbolical because the path - ordered quantity is not easy to evaluate explicitly .
some important physical quantities , however , can be calculated in terms of @xmath6 .
one example is the baryon number current discussed in sec.i and sec.iii .
the isospin current touched upon in sec.i is also calculated through @xmath6 .
another physical quantity which can be calculated from @xmath6 is the energy - momentum tensor @xmath324 defined by @xmath325 it is straightforward to obtain a result such as @xmath326.\end{aligned}\ ] ] for the solution of this paper , we see that @xmath327 is given as a linear function of @xmath328 .
we can also obtain @xmath329\end{aligned}\ ] ] and find that it contains @xmath330 ^ 2 $ ] .
up to now , we have been considering lightlike momenta @xmath331 and @xmath65 .
we note that @xmath332 should be non - negative for all pairs of future directed timelike vectors @xmath174 and @xmath64 @xcite .
the solutions obtained in this paper are of wave character .
they are regarded as the superpositions of three plane waves .
we end this paper by asking what the _ ansatz _ leading to the exact solitonic solutions of the skyrme model is .
the authors thank shinji hamamoto , takeshi kurimoto , hitoshi yamakoshi , hiroshi kakuhata , kouichi toda , masataka ueno , makoto nakamura , and hideaki hayakawa for discussions .
one of the authors ( m.h ) is grateful to noriaki setoh and takashi sugatani for valuable comments .
this work was supported in part by a japanese grant - in - aid for scientific research from the ministry of education , culture , sports , science and technology ( no.13135211 ) .
99 t. h. r. skyrme , nucl . phys .
* 31 * , 556 ( 1961 ) .
g. s. adkins , c. r. nappi , and e. witten , nucl .
* b 228 * , 552 ( 1983 ) .
a. d. jackson and m. rho , phys .
51 * , 751 ( 1983 ) .
y. m. cho , phys . rev
. lett . * 87 * , 252001 ( 2001 ) .
l. faddeev and a. j. niemi , nature ( london ) * 387 * , 58 ( 1997 ) . m. hirayama and j. yamashita , phys . rev . *
d 66 * , 105019 ( 2002 ) .
j. gladikowski and m. hellmund , phys . rev .
* d 56 * , 5194 ( 1997 ) .
j. hietarinta and p. salo , phys . lett .
* b 451 * , 60 ( 1999 ) .
r. a. battye and p. m. sutcliffe , phys .
81 * , 4798 ( 1998 ) .
r. a. battye and p. m. sutcliffe , phys .
. lett . * 79 * , 363 ( 1997 ) . g. w. gibbons , hep - th/0302149 . | a class of exact solutions of the skyrme model is obtained .
the solutions are described by the weierstrass @xmath0 function or the jacobi elliptic function .
they are not solitonic but of wave character .
they supply us with examples of the superposition of three plane waves in the skyrme model . |
methods based upon coherent states @xcite have proved to be widely useful in investigating the ground state properties of algebraic models . in algebraic models , the hamiltonian and all physical operators
are constructed from the elements of a lie algebra , usually from a bosonic realization of @xmath3 .
such models have been applied extensively to the spectroscopy of many - body systems , including nuclei @xcite and molecules @xcite .
the coherent states of an algebraic model are obtained by repeated action of a general linear combination of boson creation operators on the vacuum state .
as variational trial states , the coherent states allow the estimation of the ground state energies and properties , yielding results which become exact in the infinite boson number limit @xcite .
they are also essential in defining the classical limit for the model @xcite , providing the geometric coordinates or dynamical variables of the model .
coherent states may also be used to study the intrinsic excitation modes of an algebraic system @xcite , through the construction of coherent states orthogonal to the ground state . however , application of this method has generally been limited to excited states involving only one intrinsic excitation quantum , or at most two @xcite , due to the complexity of calculating the necessary matrix elements . in the present work , a general expression is obtained for the matrix element of an @xmath0-body operator between coherent states constructed from multiple orthogonal linear combinations of boson creation operators .
this allows the coherent state formalism to be applied to states possessing an arbitrarily large number of intrinsic excitation quanta .
the necessary definitions for the basic ground state ( condensate ) coherent state are first presented , and the results for expectation values with respect to this state are reviewed ( section [ seccondensate ] ) . the general result for excited coherent states
is then established ( section [ secmulti ] ) . as a simple illustration ,
calculations are carried out for excited states in the @xmath4 dynamical symmetry limit of the molecular two - dimensional vibron model @xcite ( section [ secvibron ] ) .
an algebraic model based upon a bosonic realization of @xmath3 is obtained by defining @xmath5 bosonic creation operators @xmath6 , @xmath7,@xmath8 , @xmath9 obeying the canonical commutation relations @xmath10\<=0 $ ] , @xmath11\<=\delta_{i , j}$ ] , and @xmath12\<=0 $ ] .
the set of all possible bilinears of a creation and an annihilation operator then forms a basis @xmath13 , @xmath14 , @xmath8 , @xmath15 for the algebra .
the physical operators of the model , such as the hamiltonian , the angular momentum , and transition operators , are constructed as polynomials in the @xmath16 .
these operators act on the states of the fock space created by @xmath6 , @xmath7 , @xmath8 , @xmath9 .
the condensate coherent state is defined in terms of the `` condensate boson '' creation operator , which is a general linear combination @xmath17 with complex coefficients satisfying the normalization convention @xmath18 .
the normalized condensate coherent state is then @xmath19 this state is an eigenstate of the total number operator @xmath20 .
the expectation value of a one - body or two - body operator with respect to the condensate ( [ eqncoherentc ] ) was deduced by van isacker and chen @xcite , using arguments based upon formal differentiation . here
let us derive an explicit result for the expectation value of an arbitrary @xmath0-body operator , since the results of section [ secmulti ] can be obtained as a natural extension .
first , for an annihilation operator @xmath21 , note that @xmath22\<=\alpha_r$ ] , from which the relation @xmath23=n\alpha_r(b_c^\dagger)^{n-1}\ ] ] follows by the product rule for commutators .
the action of @xmath21 on the condensate is thus @xmath24 repeated application yields the expectation value of an arbitrary @xmath0-body operator , @xmath25 where the underlined superscript indicates the falling factorial @xcite , defined by @xmath26 .
for the study of excited states , it is necessary to consider coherent states which are orthogonal to the condensate state .
these are constructed using multiple different coherent boson species @xmath27 ( @xmath28 ) , defined as linear combinations @xmath29 of the basic boson creation operators . ( the procedure for obtaining values of the coefficients @xmath30 appropriate to a given model is discussed in , _
e.g. _ , ref .
@xcite . ) the coefficients @xmath30 are chosen to obey the orthonormalization convention @xmath31 .
consequently , the coherent bosons satisfy canonical commutation relations @xmath32\<=0 $ ] , @xmath33\<=\delta_{s',s}$ ] , and @xmath34\<=0 $ ] .
the different @xmath27 represent the ground state condensate boson and one or more orthogonal excitation modes .
the normalized multi - species coherent state is @xmath35 the coherent state is an eigenstate of the total number operator , of eigenvalue @xmath36 .
states of different coherent boson occupation numbers @xmath37 , @xmath8 , @xmath38 are orthogonal .
the matrix element of an arbitrary @xmath0-body operator with respect to two multi - species coherent states can now be deduced following the approach of section [ seccondensate ] .
the commutation relations generalize to @xmath39\<=\alpha_{s , r}$ ] and @xmath40 = \sum_{t=1}^s n_t \alpha_{t , r}\bigl [ \prod_{s=1}^s(b_s^\dagger)^{n_s-\delta_{s , t } } \bigr].\ ] ] the action of @xmath21 on the multi - species coherent state is thus @xmath41 this is analogous to the result of equation ( [ eqnactionc ] ) , but with a separate term arising from the action of @xmath21 on each species of coherent boson contributing to the coherent state .
it is useful to define a counting function @xmath42 , giving the number of the @xmath43 which are equal to @xmath44 .
then , for a product of annihilation operators acting on the coherent state , @xmath45 \bigl ( \prod_{i=1}^m \alpha_{t_i , r_i } \bigr ) \\ \times \bigl| \bigl(n_1-\nu(t_1,\ldots , t_m;1)\bigr)\cdots\bigl(n_s-\nu(t_1,\ldots , t_m;s)\bigr)\bigr\rangle . } \ ] ] the matrix element of an arbitrary @xmath0-body operator ( @xmath46 ) with respect to two arbitrary multi - species coherent states is the inner product of two such expressions , @xmath47 \bigl ( \prod_{i=1}^m \alpha_{t_i',r_i'}^ * \alpha_{t_i , r_i } \bigr ) , } \ ] ] where the abbreviations @xmath48 and @xmath49 have been used .
three stages are involved in evaluating the matrix element of a general operator : reexpression of the operator in terms of normal - ordered @xmath0-body terms , evaluation of the matrix elements of these by equation ( [ eqnmemulti ] ) , and simplification of the result . for complicated operators or
if many coherent boson species are involved , these steps can most effectively be carried out though computer - based symbolic manipulation .
a few useful special cases of equation ( [ eqnmemulti ] ) are summarized in the appendix .
the multiple sum in equation ( [ eqnmemulti ] ) nominally contains @xmath50 terms .
however , for fixed numerical values of the @xmath51 and @xmath52 , many of the terms vanish identically due to the restriction on indices imposed by the product of kronecker delta symbols .
a summary of the number of nonvanishing terms for various @xmath53 and @xmath0 is given in table [ tabterms ] .
a given species @xmath44 of coherent boson is overannihilated when @xmath54 or @xmath55 , yielding a vanishing falling factorial in equation ( [ eqnmemulti ] ) .
thus , additional terms vanish if the expression is evaluated for a small value ( @xmath56 ) of any of the @xmath51 or @xmath52 , as typically occurs when the lowest - lying excited states are considered .
if the multiple sum in equation ( [ eqnmemulti ] ) is instead to be evaluated with the @xmath51 and @xmath52 retained as variables , all terms involving the same product of falling factorials may be collected .
this product is identical for terms with the same values of all the @xmath57 and @xmath58 .
since @xmath59 and @xmath60 , and similarly for @xmath58 , the number of distinct terms after collection is the square of the number of possible partitions of @xmath0 over @xmath53 bins .
@rrrrr & + & + @xmath53 & @xmath61-body & @xmath62-body & @xmath63-body & @xmath64-body + 2 & 2 & 6 & 20 & 70 + 3 & 3 & 15 & 93 & 639 + 4 & 4 & 28 & 256 & 2716 + 5 & 5 & 45 & 545 & 7885 + multi - species coherent states of the form ( [ eqncoherentmulti ] ) are also encountered as the condensate states of systems involving multiple constituents , each separately conserved .
an example from nuclear physics is the proton - neutron interacting boson model ( ibm-2 ) @xcite , in which proton pairs ( created by @xmath65 , @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 ) and neutron pairs ( created by @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 ) are separately conserved .
the physical operators are constructed from the elements of a lie algebra @xmath77 , and a condensate state with good boson number for each constituent is constructed as @xmath78 . since the condensate bosons @xmath79 ( @xmath80 ) are constructed from disjoint sets of boson operators , the expectation value of an @xmath0-body operator in general factorizes into the product of simple expectation values of the type ( [ eqnmec ] ) ( _ e.g. _ , ref .
* ( c1 ) ) ) .
however , the general result ( [ eqnmemulti ] ) for the multi - species coherent state matrix element can provide the simplest framework for computer - based symbolic evaluation @xcite .
the two - dimensional vibron model @xcite is the @xmath1 algebraic model , describing a system containing a dipole degree of freedom constrained to planar motion .
the basic example of such a system is a triatomic linear bender molecule , but the model is easily extended to more complex molecular systems .
the @xmath1 algebra is realized in terms of the three bosonic operators @xmath81 , @xmath82 , and @xmath83 , which satisfy canonical commutation relations .
it is convenient to define circular bosons @xmath84
@xcite , such that the operators @xmath81 , @xmath85 , and @xmath86 carry @xmath87 , @xmath88 , and @xmath89 units of two - dimensional angular momentum .
the physical operators include the angular momentum operator @xmath90 , the dipole operators @xmath91 , and the quadrupole operators @xmath92 .
the @xmath1 algebra contains the subalgebra chains @xcite @xmath93 the dynamical symmetry associated with the @xmath94 chain yields spectra matching those of the cylindrical oscillator ( pschl - teller potential ) , while the dynamical symmetry associated with the @xmath4 chain yields spectra like those of the displaced cylindrical oscillator ( morse potential ) .
the @xmath4 limit is used in the following illustrations , as the less trivial case .
the @xmath4 subalgebra is spanned by @xmath95 and has quadratic casimir operator @xmath96 .
the @xmath97 subalgebra is simply the two - dimensional angular momentum algebra , containing @xmath98 .
the simplest hamiltonian with @xmath4 dynamical symmetry is @xmath99 , which has eigenvalues @xmath100,\ ] ] with @xmath101 and @xmath102 [ figure [ figcombo](a ) ]
. dynamical symmetry of the two - dimensional vibron model , with @xmath103 .
( a ) schematic @xmath4 level energy diagram , with arrows indicating the dipole ( solid ) and quadrupole ( dashed ) transitions considered in panels ( c , d ) .
( b ) energies of all @xmath4 representations .
( c ) intensity of the @xmath104 dipole transition within each representation .
( d ) intensity of the @xmath105 quadrupole transition within each representation ( solid curve , filled circles ) and of the @xmath105 and @xmath106 @xmath107 transitions between representations ( dashed curve , open circles ) . an @xmath108 energy splitting is included in panel ( a ) , to provide visual separation of levels within an @xmath4 representation .
all observables are plotted rescaled by @xmath109 . ]
the condensate boson and an orthogonal excitation boson for the two - dimensional vibron model may be defined as @xmath110 ( see ref .
@xcite for further discussion of the choice of boson operators in the vibron model . )
the general excited coherent state is @xmath111^{n - n_x } [ b_x^\dag(r)]^{n_x } \lvert0\rangle.\ ] ] the expectation value of the @xmath4 casimir operator with respect to an arbitrary excited coherent state is evaluated using equation ( [ eqnmetwospeciestwobody ] ) , yielding @xmath112+\frac{4}{(1+r^2)^2}[n(n-1)-6n_x(n - n_x)]r^2 . } \ ] ] the equilibrium value of @xmath113 is found by minimization of the variational energy @xmath114 , giving @xmath115 . with this value of @xmath113 ,
the excited coherent state energies are @xmath116.\ ] ] if the intrinsic excitation number @xmath117 is identified with the @xmath4 quantum number @xmath118 , comparison of equations ( [ vibronenergyso3 ] ) and ( [ vibronevwso3 ] ) shows that these expressions differ only by a term of order @xmath119 .
thus , to leading order in @xmath119 , the coherent state estimate reproduces the excitation energies for all excited states , as illustrated in figure [ figcombo](b ) .
transition strengths may be estimated using coherent states , from the squared matrix element of the transition operator , @xmath120 .
the coherent state @xmath121 is not an angular momentum eigenstate , so the resulting estimate for the transition strength between two intrinsic excitations is effectively averaged over the many states of different angular momenta constituting that excitation .
this coherent state estimate thus can not be expected to provide the exact transition intensity between any two particular angular momentum eigenstates .
it does , however , indicate the general magnitude of the transition strengths and the overall dependence on excitation quantum number , and it can be quantitatively accurate if the angular momentum dependence of transition strengths is weak .
( alternatively , angular momentum eigenstates may be projected from the coherent states @xcite , but this requires additional machinery beyond simple evaluation of an @xmath0-body operator matrix element . ) dipole infrared transitions in the two - dimensional vibron model are , to leading order , induced by the operators @xmath122 , and quadrupole raman transitions are induced by the operators @xmath123 @xcite .
the strengths of transitions within an intrinsic excitation are estimated from the expectation values @xmath124 and those between succesive intrinsic excitations from @xmath125 the coherent state estimates of transition intensities for the @xmath4 dynamical symmetry are obtained from these equations with @xmath115 .
the estimates are plotted for @xmath103 in figure [ figcombo](c , d ) , together with exact values obtained by numerical diagonalization , as functions of the excitation quantum number @xmath118 ( or @xmath117 ) .
the coherent state estimate for dipole transitions closely reproduces the strength of the angluar momentum @xmath126 transition within an @xmath4 representation [ figure [ figcombo](c ) ] .
dipole transitions between different representations are forbidden , and the coherent state estimate indeed vanishes .
quadrupole transition strengths exhibit greater angular momentum dependence within a representation , and the coherent state estimate is consequently less accurate .
the strengths of quadrupole transitions involving the low angular momentum members of the representations are reproduced to within @xmath127 , except at the highest intrinsic excitation quantum numbers [ figure [ figcombo](d ) ] .
note that the angular momentum @xmath128 and @xmath129 transitions between two representations ( @xmath130 ) differ in strength , and the coherent state estimate consistently behaves as their average .
the present results serve as a basis for application of the coherent state formalism to states with an arbitrary number of intrinsic excitation quanta . this process yields estimates of eigenvalues and operator matrix elements for excited states valid to leading order in @xmath119 .
the illustration provided was to a dynamical symmetry limit of a simple model , but the coherent state analysis will likely be most useful when applied to transitional hamiltonians , for which analytic results are not otherwise available .
the coherent state formalism has in the past provided not only a quantitative calculational tool but , perhaps more importantly , a method for obtaining qualitative understanding of the equilibrium properties and fundamental modes of a system .
most , if not all , of the raw numerical results of the coherent state formalism can also be obtained by numerical diagonalization .
it is thus the latter , interpretational aspects of the coherent state formalism , and the explicit analytic forms obtained for the parameter dependences of observables , which have proved most useful .
the present results for multiply excited states thus might most productively be used in investigating the general nature of the evolution of a system s properties with excitation energy .
discussions with f. iachello , a. leviatan , and f. prez - bernal are gratefully acknowledged .
this work was supported by the us doe under grant de - fg02 - 91er-40608 .
some useful special cases of the general matrix element ( [ eqnmemulti ] ) are given here .
if the coherent states involve only two species of coherent boson ( @xmath131 ) , then the 1-body and two - body operator matrix elements are @xmath132 and @xmath133 with all others zero . for an expectation value ( all @xmath134 ) ,
equation ( [ eqnmemulti ] ) simplifies to @xmath135 \bigl ( \prod_{i=1}^m \alpha_{t_i',r_i'}^ * \alpha_{t_i , r_i } \bigr ) . } \ ] ]
10 url # 1#1urlprefix[2][]#2 zhang w , feng d h and gilmore r 1990 _ rev .
phys . _ * 62 * 867 the definitions @xmath136
@xcite are used , rather than @xmath137 @xcite .
this choice is necessary for @xmath122 to be the physical dipole operators and provides a closer correspondence with the treatment of the 3-dimensional vibron model @xcite . | a general expression is obtained for the matrix element of an @xmath0-body operator between coherent states constructed from multiple orthogonal coherent boson species .
this allows the coherent state formalism to be applied to states possessing an arbitrarily large number of intrinsic excitation quanta . for illustration ,
the formalism is applied to the two - dimensional vibron model [ @xmath1 model ] , to calculate the energies of all excited states in the large-@xmath2 limit . |
the computational power of quantum computers has threatened classical cryptosystems . for example ,
public key cryptosystems , such as rivest - shamir - adleman public key cryptosystem @xcite , can be broken by quantum computers to be able to perform the fast factorization .
on the other hand , quantum mechanical phenomena provide us a new kind of cryptosystems , called quantum key distribution ( qkd ) , from which we can in principle obtain perfectly random and secure key strings .
the first quantum cryptographic protocol was presented by bennett and brassard @xcite and their protocol bore the acronym bb84 . in 1991 ,
ekert @xcite proposed a qkd protocol using entangled particles .
it was modified by bennett , brassard , and mermin @xcite .
let us call the modified version the einstein - podolsky - rosen ( epr ) protocol .
the epr protocol is a qkd between two persons using an epr pair of spin @xmath0 particles in the state @xmath1 . using the greenberger - horne - zeilinger ( ghz ) state @xmath2 @xcite the secret sharing protocol
was presented by hillery , bu@xmath3ek and berthiaume @xcite . in this protocol
, alice distributes the information on a key to bob and charlie . and
the key can be restored only when their information are collected by them . in this paper , applying the secret sharing protocol , we generalize the epr protocol on noiseless channels by the properties of several cat states @xcite and then obtain qkd protocols between group @xmath4 and group @xmath5 . in each group the information of a secret key is distributed to all members . after the process for recovery of the key
, the two groups get the secret key . and the protocols require each member s approval and cooperation .
furthermore , when some members try to affect the shared bit adversely , if the shared key does not have the correct correlation ( or anti - correlation ) then it should be revealed to others in the test step .
any external eavesdropper should also be detected even if several members assist the eavesdropper .
this paper is organized as follows : in section 2 , we investigate some properties of several cat states .
the qkd protocol between two groups and its modification are presented in section 3 .
we analyze the security for the protocol in section 4 .
let us begin with reviewing cat states @xcite .
the t - particle cat state is defined as a entangled state of the type @xmath6 whereby @xmath7 stands for the binary variable in @xmath8 , and @xmath9 .
furthermore , equation ( [ eq : cat ] ) becomes one of the bell states when @xmath10 and one of the ghz states when @xmath11 . from now on , we use the following several cat states : @xmath12 we define @xmath13 , @xmath14 , @xmath15 , and @xmath16 . for @xmath17 , we notice the states in equation ( [ cat state1 ] ) and ( [ cat state2 ] ) have the following relations : @xmath18 when @xmath19 is a group of t persons , assume that , for one of the above four cat states , each person takes its one particle and measure in the @xmath20- or @xmath21-direction .
firstly we let @xmath22 be the number of members modulo 4 who measure in the @xmath21-direction , @xmath23 , and @xmath24 the sum of the measurement outcome of all members modulo 2 .
then the following results are obtained .
* suppose @xmath22 is even .
then @xmath25 is @xmath26 for @xmath27 , and it is @xmath28 for @xmath29 , where @xmath30 ( mod 2 ) for any @xmath31 .
* suppose @xmath22 is odd .
then @xmath25 is @xmath26 for @xmath32 , and it is @xmath28 for @xmath33 . also , it is noticed that if the above suppositions of @xmath22 are not satisfied , @xmath25 becomes 0 or 1 with probability @xmath0 i.e. it has no rules . using an induction on @xmath34 the proof of such facts is given . to begin with , for @xmath35
it is trivial .
assume that these statements are true for @xmath36 .
the cat state @xmath37 is considered .
let @xmath22 be even .
equation ( [ equ : decomposition ] ) implies @xmath38 if any one member takes measurement in the @xmath20-direction and obtains 0 then @xmath39 and @xmath40 will be even , where g@xmath41 is the group of all members except that member .
from equation ( [ equ : catphi ] ) @xmath42 and @xmath43 . thus @xmath44 .
otherwise , @xmath45 by ( [ equ : catphi ] ) and @xmath46 .
thus @xmath44 . on the other hand , for the case
that the member takes a measurement in the @xmath21-direction , the proof is similar to the above case .
hence , we hold that @xmath47 . that is , the previous assumption holds for @xmath34 . for other cat states ,
all of the proofs are similar .
now , we consider two parties , @xmath4 and @xmath5 , that consist of @xmath48 members and @xmath49 members , respectively . applying the previously described properties of the cat states , we obtain the table [ table : exencoding ] . [ cols="^,^,^,^,^,^ " , ] if @xmath50 and @xmath51 , it follows from the condition @xmath52 that the number of required test bits is not less than 270 .
however , for fixed @xmath48 and @xmath49 the fewer number of test bits are required .
the change of the error rate according to the number of test bits is exemplified in tables [ table : pro1,6,6 ] for the case that @xmath53 and @xmath54 the probability in the modified protocol has a little difference from the primary protocol because there are no discarded bits and @xmath55 is determined by @xmath56 and @xmath57 . by removing the strategy to use @xmath55 we can easily get the following error rates of the modified protocol . in the case @xmath58 , @xmath59 in the case @xmath60 , @xmath61 as in the case of the first protocol
, we analyze just the case @xmath60 .
the error rate in equation ( [ eq : pro2ab ] ) has no connections with the values of @xmath62 and @xmath48 , and depends just on @xmath63 and @xmath49 .
for @xmath64 , we require only @xmath34 that satisfies @xmath65 . for @xmath66
, it becomes @xmath67 . in the modified protocol
, we can notice that two groups require remarkably smaller test bits than the first protocol , and furthermore errors can be detected from the test step even in the cases @xmath68 or , @xmath69 and @xmath70 , while errors can not be detected for the case in the first protocol .
we remark that if just a collector can take the information on @xmath71 or only one member plays the collector for all shared bits then a fewer test bits would be required .
we assume that eve intercepts @xmath49 particles travelling from @xmath4 to @xmath5 , and call this state ` the intercepted state ' .
she chooses an @xmath72-particle cat state and resends @xmath49 particles of this cat state to @xmath5 ( @xmath73 ) .
we refer to the remainder ( @xmath74)-particle state as ` the remainder state ' . before announcement of @xmath75 ( or @xmath76 ) the measurement of the intercepted state ( or the remainder state ) can not give her the information on @xmath77 ( or @xmath78 ) .
so she should measure on the intercepted state and the remainder state , after @xmath75 and @xmath76 are announced .
even though she measures in the way , she should have the information on @xmath71 to obtain @xmath77 , since @xmath77 is completely determined by @xmath71 and @xmath75 . in order to take information on @xmath79
, she needs any conspirator in @xmath4 . on the other hand , she wants to change @xmath77 or @xmath78 into the values she desires to prevent errors from occurring .
thus she needs collectors assistance in @xmath4 or @xmath5 . without any conspirators the test induces errors with probability @xmath80 in the first protocol and probability @xmath0 in the modified protocol , respectively .
for these facts eve s strategy makes at least the following error rate in the first protocol . in the case
@xmath58 , @xmath81 in the case @xmath82 , @xmath83 in the modified protocol the error rates are similar to the first protocol , because eve can not have a different strategy . from these equations
we can know that this strategy is not optimal to eve .
applying the properties of cat states and the secret sharing @xcite , we proposed two generalized qkd protocols between two groups and showed that the protocols are secure against an external eavesdropper using the intercept / resend strategy .
the importance of these protocols is that any member in the two groups can not obtain the secret key strings without cooperation , that is , the secret key strings can be obtained only under all member s approval . *
acknowledgments * s.c . acknowledges the support from ministry of planning and budget and thanks s.lee for discussions .
d.p.c . acknowledges the support from korea research foundation ( krf-2000 - 0150dp0031 ) . | in this paper , we investigate properties of some multi - particle entangled states and , from the properties applying the secret sharing present a new type of quantum key distribution protocols as generalization of quantum key distribution between two persons . in the protocols each group can retrieve the secure key string , only if all members in each group should cooperate with one another .
we also show that the protocols are secure against an external eavesdropper using the intercept / resend strategy . |
fundamental quantum phenomena , such as non - locality and entanglement of quantum degrees of freedom , have regained a lot of interest recently , mainly due to their potential usefulness as a computational resource .
mesoscopic physics provides a new platform for the investigation of these phenomena , important issues being the creation , quantification , and verification of non - locality / entanglement . in this paper
, we describe an experiment where two electrons with different orbital wave functions are superposed in an interferometer and analyzed in a bell type experiment involving two - particle correlation measurements , see fig .
[ fig : setup_a ] .
the particular feature of this bell test is the replacement of the four different settings of local detectors in the original setup by four different time - delays in the measured correlators .
the main physical property we want to exploit is the indistinguishability of quantum particles , which results in a specific interference between different trajectories .
we wish to convey three messages : first , the non - locality in the time - delayed correlations due to indistinguishability manifests itself in the violation of a bell inequality .
second , the degree of violation is related to the accuracy of the measurement and is reduced , once the local measurement can distinguish between the different orbital wave functions of the particles .
the above two items refer to spinless objects ( or particles with equal spin ) .
third , adding the spin degree of freedom , we show how the symmetry relation between spin- and orbital components allows to extract information on spin - entanglement from an orbital measurement . and @xmath0 with wave functions @xmath1 and @xmath2 are mixed in a four - terminal splitter ( characterized by the mixing angle @xmath3 ) and analyzed at @xmath4 through measurement of time - correlations during the time intervals @xmath5 and @xmath6 ( time axis drawn perspectively into the plane ) .
we are interested in the entanglement of the lead indices u and d with respect to bipartitioning of the system between the time intervals @xmath7 and @xmath8 . for details on the implementation of the source ( shaded area )
see fig .
[ fig : source].,width=283 ] by now , numerous proposals have been made how to create entangled states in mesoscopic setups , both for orbital- and spin degrees of freedom @xcite .
the verification and quantification of entanglement can be carried out using bell inequality checks @xcite or state tomography @xcite .
the indistinguishability of ( spinless ) particles producing a two - particle aharonov - bohm effect and entanglement has been exploited in a hanbury - brown twiss interferometer @xcite ; here , we offer an alternative implementation which makes use of an electron beam splitter . using the same setup as discussed here , burkhard _
et al._@xcite have demonstrated how to distinguish between singlet and triplet spin - states by a measurement of zero - frequency cross - correlations .
below , we exploit that the orbital measurement of the bell parameter preserves the spin - entanglement ; this feature allows us to find a lower bound on the concurrence of the spin wave function . in the following ,
we consider a setup with two incoming leads , denoted as @xmath9 and @xmath10 , connected to two outgoing leads @xmath11 and @xmath12 through a reflectionless four - terminal beam splitter , see fig .
[ fig : setup_a ] . at time @xmath13
two electrons with normalized orbital wave functions @xmath1 and @xmath2 and common spin state @xmath14 are injected into the leads @xmath15 and @xmath10 .
the state , factorizable in orbital and spin parts ( with factorized orbital part and general spin part ) , is conveniently written within a second quantized formalism , @xmath16 here , @xmath17 creates electrons at the position @xmath18 in lead @xmath19 and @xmath20 is the vacuum state with no electrons . after mixing in a four - terminal splitter ( with mixing angle @xmath3 ) , we will analyze correlations in the system through detection of particles in time ( see fig.[fig : setup_a ] ) or space separated intervals @xmath7 and @xmath8 ( see fig .
[ fig : setup_b ] ) .
our focus then is on the entanglement of the lead indices u and d with respect to bipartitioning of the system between the time or space intervals @xmath7 and @xmath8 .
the particles in the outgoing leads @xmath11 and @xmath12 are subjected to a bell test expressed through time - resolved current - current correlators in the leads @xmath21 and @xmath22 , @xmath23 ( both auto- @xmath24 and crossed- @xmath25 correlators are considered ) , @xmath26 where @xmath27 is the total current operator ( summed over spin degrees of freedom ) in lead @xmath19 at position @xmath18 and time @xmath28 .
the time integration is taken over a finite time interval @xmath29 $ ] ( same for @xmath8 ) with the width @xmath30 accounting for the finite time - resolution of the current measurement , cf .
fig.[fig : setup_a ] ; the limit @xmath31 corresponds to a measurement of the instantaneous current . in the following
, we will assume that all correlators are measured at some fixed symmetric position @xmath32 and omit the coordinate variable . with only two electrons present in the system and for non - overlapping time - intervals @xmath33 ,
the correlation function @xmath34 is proportional to the joint probability @xmath35 for the detection of two particles during the time intervals @xmath7 and @xmath8 in the leads @xmath21 and @xmath22 , see ref .. there are four distinct possibilities to distribute two electrons between the outgoing leads and we can define the properly normalized ( @xmath36 ) probabilities as @xmath37 out of these , we define the two - particle bell inequality in the clauser - horne @xcite form in the same way as it is done in the usual optics context @xcite : we introduce the bell correlation functions @xmath38_{ab } \label{bcorr}\ ] ] and obtain the bell inequality @xmath39 here , the polarizations @xmath40 in the optics context are replaced by the lead indices @xmath11 and @xmath12 and the role of the four different polarization settings of the detectors is played by four different time intervals @xmath7 , @xmath8 and @xmath41 , @xmath42 .
the violation of this inequality for a particular choice of time intervals shows that non - local correlations are present in the system , i.e. , the result of the measurement can not be simulated by any local - variable theory .
let us demonstrate that the above bell inequality indeed can be violated by the incoming state ( [ inst ] ) after proper projection .
we then have to calculate the four current - current auto- and cross - correlators @xmath43 with @xmath44 and @xmath45 , respectively .
this is done within the scattering matrix approach to quantum noise @xcite : we assume that the fourier components @xmath46 and @xmath47 of the single - particle wave functions are concentrated near the wave vector @xmath48 , allowing us to linearize the energy
momentum dispersion near @xmath49 .
the time evolution of the incoming state ( [ inst ] ) then is described by the propagation of the single - particle wave packets @xmath1 and @xmath2 with constant velocity @xmath50 to the right , @xmath51 and @xmath52 , where @xmath53 is a retarded variable . with the scattering matrix of the beam splitter ( parametrized by the angle @xmath3 ) , @xmath54 we can express the current operators @xmath27 in the outgoing leads @xmath55 through the electronic scattering states . averaging the product of current operators in eq .
( [ corr ] ) over the incoming state @xmath56 one arrives at the results @xmath57\bigr\ } , \nonumber \\ & & \langle \hat i_\mathrm{u}(\xi_1 ) \hat i_\mathrm{d } ( \xi_2 ) \rangle = ( ev_0)^2\bigl\{\cos^4\theta\ , |f(\xi_1)|^2 |g(\xi_2)|^2 \label{cross } \\ & & \qquad\qquad\qquad\qquad\qquad + \sin^4\theta\,|g(\xi_1)|^2 |f(\xi_2)|^2 \nonumber\\ & & \quad+ \sin^2\theta \cos^2\theta\,q\,\bigl(f(\xi_1 ) g^*(\xi_1 ) g(\xi_2 ) f^*(\xi_2 ) + c.c .
\bigr)\ } , \nonumber\end{aligned}\ ] ] where @xmath58 describes the overlap between the spin states of the two electrons in the incoming state ( [ inst ] ) .
the two other correlation functions @xmath59 and @xmath60 are obtained by exchanging @xmath61 and @xmath62 in eqs .
( [ auto ] ) and ( [ cross ] ) . substituting these expressions for the current correlators into eq .
( [ corr ] ) and integrating over ( non - overlapping ) time intervals @xmath7 and @xmath8 , one arrives at the bell correlation function eq .
( [ bcorr ] ) @xmath63 where we have introduced the particle densities @xmath64 and @xmath65 averaged over the time intervals @xmath7 and @xmath8 , @xmath66 ( and similarly for @xmath65 with @xmath67 replaced by @xmath68 ) .
the overlap @xmath69 between different single particle wave functions reads @xmath70 in the following , we apply the result ( [ bcorr2 ] ) first to spinless fermions and plane wave states @xmath67 and @xmath68 and confirm the violation of the bell inequality in this simple situation .
we then proceed with a rederivation of the expression ( [ bcorr2 ] ) with space - like separated measurement intervals @xmath7 and @xmath8 in order to make the origin of the entanglement more transparent .
a formulation in terms of reduced density matrices leading to an expression of the bell correlator in terms of concurrences completes the discussion .
we first concentrate on _ spinless particles _ ; this situation can be realized by preparing the two electrons in equal spin - states , @xmath71 with corresponding overlap @xmath72 . to begin with ,
we choose a plane wave form for the wave packets with different momenta @xmath73 and @xmath74 close to @xmath49 ( in order to allow for the linearized spectrum ) , @xmath75 and @xmath76 .
the correlation function @xmath77 takes the form @xmath78 where @xmath79 is the relative phase shift accumulated by the two waves between the two measurement intervals and @xmath80 is the frequency mismatch between the two plane waves .
the phase shift @xmath81 replaces the angle between the two polarizers in the conventional bell setup .
the visibility factor @xmath82 accounts for the width @xmath30 of the time interval , @xmath83 the other correlation functions involving intervals @xmath41 and @xmath42 are obtained in the same way ; their combination into the bell inequality eq.([bi ] ) produces a maximal violation for the angles @xmath84 and @xmath85 , corresponding to measurement intervals with relative distance @xmath86 , @xmath87 , and @xmath88 , where @xmath89 and @xmath90 is an arbitrary reference time . in this case
the bell inequality eq .
( [ bi ] ) reduces to @xmath91 for given @xmath92 , the maximal violation @xmath93 is reached for a symmetric beam splitter with @xmath94 , cf .
[ fig : violation ] .
furthermore , the maximally allowed degree of violation @xmath95 can be attained only for @xmath96 , corresponding to a short time measurement of the current value with @xmath97 : hence , the maximal violation of the bell inequality can be obtained for indistinguishable particles , while a time interval with length beyond heisenberg s uncertainty bound @xmath98 allows for a distinction between the two particles and the bell inequality can not be violated in this classical situation .
for @xmath99 the bell inequality is always satisfied . within the region
@xmath100 , the bell inequality ( [ bi2 ] ) is always violated for any mixing angle @xmath101 , although to a lesser degree then in the symmetric point @xmath102 . versus mixing angle @xmath3 ( solid line : spin - triplet states ;
dashed line : spin - singlet state).,width=283 ] let us discuss the physical origin of the violation .
the correlator eq.([corr ] ) measured in the bell test is finite , provided that both electrons are detected within the time windows @xmath7 and @xmath8 ; in this case , it is proportional to the probability @xmath103 .
although , formally , the electrons have different energies @xmath104 and thus are distinguishable in principle , given a small time resolution @xmath105 of the local current measurements one can not distinguish between the energies @xmath106 and @xmath107 . under this circumstances
the electrons indeed can be considered as indistinguishable particles .
then , according to the rules of quantum mechanics , there are two quantum alternatives contributing to a coincident detection of the electrons in @xmath7 and @xmath8 : either the electrons with energies @xmath108 and @xmath109 are detected in the time windows @xmath7 and @xmath8 , respectively , or vice a versa .
these two alternatives contribute to the measurement outcome with different phases : in the first case , the phase factor acquired by the two - particle wave function after the first measurement at @xmath90 due to the propagation of the second particle until @xmath110 is given by @xmath111 $ ] , while in the second case this phase assumes the value @xmath112 $ ] .
the phase difference between the two alternatives leads to quantum interference and a corresponding oscillatory dependence ( with frequency @xmath113 ) of the probability @xmath114 as a function of time , with an amplitude proportional to the visibility factor @xmath92 .
the precise bound on @xmath30 allowing for a violation of the bell inequality is given by @xmath115 or @xmath116 , corresponding to a measurement where the heisenberg uncertainty principle for energy
time variables is violated . in order to understand better the nature of the entanglement observed in ( [ bi2 ] ) , we consider a slightly different experiment , where instead of using time - separated detection intervals , the two observers alice and bob are measuring the simultaneous appearance of particles in spatially separated regions @xmath7 and @xmath8 of the setup , see fig .
[ fig : setup_b ] ; for particles with a linear dispersion , these two experiments are equivalent since a time delayed measurement with @xmath117 at the point @xmath18 corresponds to a coincident measurement at time @xmath28 with @xmath118 .
we first concentrate on plane - wave incoming states , where the present setup with spatially separated detectors provides additional insights .
in particular , we will see that it is the projection of the non - entangled incoming state onto the two domains @xmath7 and @xmath8 that defines a bipartition of the system with respect to which the lead index becomes entangled . on the other hand , in order to perform a bell inequality check , we need a set of local ` rotations ' of the measurement apparatus : in our setup , the parameters generating a suitable set of local ` rotations ' are determined by the distance between the measurement domains @xmath7 and @xmath8 and by the mixing angle @xmath3 .
a central element in our discussion below is the interchangeability of mixing @xmath119 and projection @xmath120 onto the domains @xmath7 and @xmath8 , where @xmath121 denotes the one - particle scattering matrix of the beam splitter and the tensor product @xmath119 acts on our two - particle state .
this interchangeability is a trivial consequence of these two operations affecting different degrees of freedom , coordinates @xmath122 and @xmath123 and lead indices u and d. in terms of these operators , we can relate the incoming and outgoing states via @xmath124 assuming that the projections onto @xmath7 , @xmath8 in the outgoing leads and onto @xmath125 , @xmath126 in the incoming leads ( see fig .
[ fig : setup_b ] ) are ballistically separated ( i.e. , the measurement in @xmath7 , @xmath8 involves the appropriate ballistic delay time ) we can write @xmath127 hence , in our discussion we are free to interchange the two operations of mixing and projection .
consider then an incoming state ( before mixing ) with single - particle wave functions @xmath128 and @xmath129 with shifted momenta .
the state incident from leads @xmath15 and @xmath0 can be written as a simple slater determinant , @xmath130 and thus is non - entangled . the lead index @xmath131 of the electron field operator @xmath132 is conveniently regarded as a pseudo - spin . and @xmath0 with wave functions @xmath1 and @xmath2 are mixed in a four - terminal splitter and analyzed through measurement of equal - time correlations within the space intervals @xmath7 and @xmath8 centered around @xmath133 and @xmath134 .
the interchangeability of projection ( to the intervals a and b ) and mixing allows to shift the measurement intervals to the positions @xmath125 and @xmath126 in the incoming leads ; provided that the measurements in @xmath125 and @xmath126 and in @xmath7 and @xmath8 are ballistically delayed in time , the measurement outcome is the same.,width=283 ] to start with , we analyze the coincident detection of two particles within the non - overlapping regions @xmath135 and @xmath136 of the incoming leads , see fig.[fig : setup_b ] , and select only those events , where each of the observers ( alice in @xmath135 and bob in @xmath136 ) finds only one particle . for two particles ,
this projection can be described by the operator @xmath137 , with the particle number operator @xmath138 counting particles in the region @xmath139 of the incoming leads . projecting the incoming state ( [ 2wave ] )
one arrives at the state @xmath140\ , \hat\psi_{\bar\mathrm{u}}^\dagger(x_1 ) \hat\psi_{\bar\mathrm{d}}^\dagger(x_2)|0\rangle , \nonumber\end{aligned}\ ] ] where @xmath141 and @xmath142 are equal to @xmath1 and @xmath2 for @xmath143 and vanishing outside .
this projected state is no longer a simple slater determinant and describes a two - particle state entangled in the lead indices and shared between the regions @xmath125 and @xmath126 of the incoming leads .
it is instructive to rewrite the state ( [ 2prj ] ) in a pseudo - spin notation : assuming for simplicity that the intervals @xmath135 and @xmath136 are reduced to individual points @xmath144 and @xmath145 we have @xmath146 where @xmath147 and @xmath148 denote states of particles localized in @xmath139 and residing in lead @xmath15 and @xmath0 , respectively ; the orbital part of the wave function contributes the phase factors @xmath149 with @xmath150 , where @xmath151 is a momentum mismatch .
the projected state ( [ 2prj_s ] ) is in fact maximally entangled in the lead- or pseudo - spin index with respect to bipartitioning the system between the regions @xmath135 and @xmath136 . in the following ,
we wish to detect this entanglement in a bell test .
the implementation of a bell test relies on the ability to locally change the pseudo - spin basis of the particles . to do so
, we transmit the original incoming state eq .
( [ 2wave ] ) through a beam splitter before measuring the presence of particles in the intervals @xmath7 and @xmath8 , now located in the leads u and d to the right of the mixer , see fig . [
fig : setup_b ] ; the mixing then acts as an equal rotation of the ( pseudo - spin ) basis @xmath152 for both particles .
however , such a global rotation of the original basis alone is not sufficient to perform the bell test , as _ locally distinct _ rotations are required as well ; the latter are implemented through different choices in the separation @xmath153 between the regions @xmath7 and @xmath8 . exploiting the interchangeability of projection and mixing , cf .
( [ eq : interch_1 ] ) and ( [ eq : interch_2 ] ) , we see that this change in distance results in a relative rotation with the angle @xmath154 around the original @xmath155-polarization axis of the pseudo spins , see eq .
( [ 2prj_s ] ) . writing the outgoing state ( [ eq : interch_2 ] ) in pseudo - spin notation
, we obtain the expression @xmath156 this projected state describes two spatially separated localized particles with entangled pseudo spin indices . choosing different space separations between the regions @xmath7 and @xmath8 allows one to change the phase @xmath81 and mixing by @xmath119 generates a second rotation parametrized by the angle @xmath3 . calculating the joint probabilities @xmath157 for the four settings @xmath158 we find the bell correlation functions @xmath77 as given by ( [ bcorr2 ] ) and choosing appropriate angles @xmath159 and @xmath3 one finds the bell inequalities violated . in a last step ,
we reformulate our analysis in terms of density matrices and express the bell inequality in terms of concurrences of density matrices reduced after projection to the intervals @xmath7 and @xmath8 .
we rewrite the projected state ( [ 2prj ] ) incident from leads @xmath15 and @xmath0 in pseudo spin representation , @xmath160 the joint measurement of the pseudo - spin index in the regions @xmath135 and @xmath136 is described by the coordinate - reduced two - particle density operator @xmath161 we introduce the two - particle pseudo - spin basis @xmath162 @xmath163 ( @xmath136 ) ; the normalized density matrix then assumes the form @xmath164 where @xmath165 , @xmath166 and @xmath167 with @xmath168 .
although initially the two particles have been in a pure state , the _ reduced _ density matrix ( [ rab ] ) corresponds to a mixed state with @xmath169 .
the calculation of the entanglement in the mixed two - particle state eq .
( [ rab ] ) corresponds to finding the concurrence @xmath170 of a two - qubit problem , and thus can be calculated following the scheme introduced by wootters @xcite , @xmath171 where @xmath172 are the eigenvalues of the matrix @xmath173 with @xmath174 , @xmath175 is a pauli matrix , and @xmath176 denotes the tensor product .
the result of this calculation provides us with the expression @xmath177 this quantity is indeed restricted to the interval @xmath178 $ ] , as follows from the cauchy - schwartz inequality and the inequality @xmath179 , @xmath180 .
the state described by eq .
( [ rab ] ) is trivial ( i.e. , not entangled or classical ) only for zero overlap @xmath181 and/or @xmath182 .
physically , the vanishing of the overlap between the wave functions @xmath1 and @xmath2 in either of the two regions @xmath125 and @xmath126 implies , that these orbital states are _ perfectly distinguishable _ via a local measurement . in this situation the corresponding density matrix @xmath183 can be written in a convex form @xmath184 , with probabilities @xmath185 and @xmath186 , and thus is separable . on the other hand , for @xmath187 one can not perfectly distinguish between the different orbital states via local measurements , resulting in an interference between the different terms of the anti - symmetric wave function eq .
( [ 2wave_1 ] ) , a non - separable density matrix , and a finite concurrence .
next , we analyze the reduced density matrix in the outgoing leads , i.e. , in the new basis @xmath188 . exploiting the possibility of exchanging real space projection and mixing
, we can simply rotate the projected density matrix according to @xmath189 the diagonal elements of the density matrix @xmath190 directly provide the detection probabilities @xmath103 , which then can be used in the calculation of the bell correlation function @xmath77 , eq .
( [ bcorr ] ) , @xmath191 where the angle @xmath81 is given by the overlap integrals , @xmath192 ; combining eqs .
( [ eq : conc ] ) and ( [ eq : eabc ] ) we immediately recover the original expression ( [ bcorr2 ] ) . choosing four different intervals
@xmath7 , @xmath8 , @xmath41 , and @xmath42 ( note that the selection of these intervals is non - trivial in the general situation discussed here , as @xmath81 now involves overlap integrals ) , we can set up the bell inequality ( [ bi ] ) and find the result expressed in terms of concurrences @xmath193 , @xmath194 + 2\cos^2(2\theta ) \bigr|\leq 2 .
\nonumber\end{aligned}\ ] ] choosing plane waves for @xmath1 and @xmath2 , the concurrences take the value @xmath195 and the bell inequality reduces to the simpler form found earlier , see eq.([bi2 ] ) ; for the general case , the degree of violation depends separately on the shapes @xmath1 and @xmath2 of the orbital wave functions in each region @xmath7 , @xmath41 , @xmath8 , and @xmath42 via the corresponding concurrences .
so far , we have considered only spinless particles or , more exactly , two electrons in a spin - triplet state with the same spin polarization of the electrons , @xmath196 and @xmath197 . since all spin - dependence of the bell inequality is encoded in the overlap @xmath198 of the spin wave - functions ,
see eq.([bcorr2 ] ) , one concludes that the above results are valid as well for the third maximally entangled triplet state , @xmath199 with @xmath72 . on the other hand ,
the character of violation is modified for the spin - singlet state @xmath200 with @xmath201 . choosing a set of optimal time intervals @xmath7
, @xmath8 , @xmath41 , and @xmath42 , the resulting bell inequality takes the form @xmath202 the main difference to the previous result for spin - triplet states is that this inequality can be violated only for a sufficiently large visibility factor @xmath100 and a beam splitter with a mixing angle @xmath3 sufficiently close to optimal , @xmath203 $ ] , where the critical angle @xmath204 is given by @xmath205 , cf .
[ fig : violation ] .
this result allows one to distinguish between triplet and singlet incoming states by measuring a bell inequality involving only orbital degrees of freedom , see also ref .. moreover , assuming that the incident electrons have opposite spin polarization , i.e. , their spin state can be written as a superposition @xmath206 , the degree of violation of the orbital bell inequality gives a lower bound on the value of the concurrence in the spin part of the wave function .
indeed , in this case @xmath207 and the maximal violation of the orbital bell inequality for @xmath208 and symmetric scattering is given by @xmath209 .
at the same time , the concurrence @xmath210 , with equality established for real @xmath19 and @xmath211 ( note that @xmath212 gives the degree of ( useful ) spin entanglement in the outgoing leads u and d ) .
hence , measuring the entanglement @xmath213 of the orbital part of the wave function ( which leaves the spin component untouched ) , provides a ( lower ) estimate of the degree of spin entanglement of the incoming state .
if the spin wave function of the incoming electrons factorizes , @xmath214 , the orbital bell inequality never can be violated since in this situation the electrons are distinguishable and thus the detection of an electron with given spin in one of the outgoing leads always allows to determine its origin . injects a singlet - pair of electrons into the lead s ; the @xmath215 four - terminal splitter distributes the particles with equal probabilities among the two leads @xmath216 and @xmath217 .
the resonances in the quantum dots select the desired energies @xmath108 and @xmath109 ; dots residing in the coulomb blockade regime inhibit the propagation of two electrons into the same lead , such that the two - particle incident state involves one particle in each of the leads @xmath15 and @xmath10.,width=283 ]
finally , we discuss the potential experimental realization of the proposed bell test . the source of spin - entangled incoming particles
can be realized with the help of a beam splitter , followed by leads with dots serving as energy filters defined through resonance levels at energies @xmath108 and @xmath109 , see fig . [
fig : source ] .
we assume that both dots reside in the strong coulomb blockade regime ; applying a single - electron voltage pulse to the source lead @xmath218 , two electrons in a singlet state are detached from the fermi see @xcite .
there is only one scattering process , involving trajectories where the electrons tunnel through different quantum dots , for which the two electrons reach the second beam splitter .
the incoming state in the leads @xmath15 and @xmath0 then is of a spin - singlet type with different energies @xmath108 and @xmath109 as defined through the dot resonances .
all other scattering processes with only one or no electrons propagating towards the second beam splitter are irrelevant as they do not contribute to the correlation measurement . a spin entangled triplet state can be generated with the help of spin - polarized reservoirs with polarizations @xmath219 and @xmath220 attached to the leads @xmath218 and @xmath221 , respectively . applying a single - electron voltage - pulse to each reservoir , two electrons with opposite spins
are injected into the leads @xmath218 and @xmath222 , see fig .
[ fig : source ] .
the state @xmath223 incident on the symmetric beam splitter emerges with a component @xmath224 |0\rangle\nonumber\end{aligned}\ ] ] describing electrons scattered into different leads @xmath216 and @xmath225 ; it is this component which can propagate through the subsequent energy filter and contribute to the current correlators .
the propagation of this component through the quantum dots results in an entangled spin - triplet state of the form given by eq .
( [ inst ] ) with @xmath226 and @xmath227 .
above , we have considered an idealized situation where only two electrons are present in the system , while in a realistic situation one deals with electronic reservoirs at finite temperature .
the associated equilibrium fluctuations then generate noise signals which are of the same order as the correlations associated with the injection of the two electrons .
we note , however , that the corresponding equilibrium current correlators @xmath228 assume significant values only for _ instantaneous _ or _ ballistically retarded _
variables , i.e. , at times @xmath229 and @xmath230 in the same leads and @xmath231 in opposite leads ( here , @xmath232 denotes the distance between the position of measurement and the reflecting dots ) .
the bell test involves correlations at time differences of the order of @xmath233 and a proper choice of the frequency mismatch @xmath234 always allows one to render the contribution from equilibrium fluctuations negligible .
another restriction on @xmath235 is due to dephasing and electron - electron interactions ; we then have to assume that the characteristic times associated with these processes are larger then @xmath235 .
we have discussed how to make use of quantum indistinguishability as a resource to generate non - classical correlations : the indistinguishability of particles enforces proper symmetrization of their wave function and results in non - factorizable states .
we have demonstrated how to generate such states with the help of quantum dots residing in the coulomb blockade regime and have determined their degree of entanglement as measured in a bell inequality test based on auto- and cross - current correlators . in a real experiment , the latter
are measured over a finite time or space domain . as a result
, we obtain an interesting interplay between the measurement accuracy ( time or space resolution ) and the degree of non - locality as measured in the bell inequality test : the more information is gained that locally distinguishes between the particles , the smaller is the degree of violation .
once the uncertainty principle allows for the identification of the particle , the bell inequality can not be violated any longer .
this feature can be exploited in the design of experiments testing the above predictions : choosing a small energy difference @xmath236 allows for a slow measurement with a less stringent time resolution , while the violation of the bell test remains observable .
on the other hand , the energy difference @xmath237 has to be chosen sufficiently large in order to avoid the influence of decoherence or interactions .
the above setup for spinless particles provides an alternative for the observation of the two - particle interference as proposed by samuelsson _
_ @xcite and recently observed by neder _
_ @xcite ; here , the role of the magnetic flux @xmath238 penetrating the hanbury - brown twiss interferometer is replaced by the time - delay of subsequent measurements in the correlator . adding the spin degree of freedom , we are confronted with two distict situations :
if the spin degree of freedom allows to distinguish between the particles ( this is the case for the spin - state @xmath239 ) the bell inequality is never violated . on the other hand ,
entangled spin states in the singlet or triplet sector ( these are the states @xmath240 and @xmath241 ) can generate maximal violation of the bell inequality ; finally , the superposition of these states reduces the spin - entanglement and the degree of violation in the orbital bell inequality gives a lower bound on the spin - concurrence , with an ideal measurement providing the best bound . c. w. j. beenakker , in _ quantum computers , algorithms and chaos _ , proceedings of the int .
school of physics enrico fermi " , varenna 2005 , vol .
* 162 * ( ios press , amsterdam , 2006 ) ; arxiv : cond - mat/0508488 . | adopting the frame of mesoscopic physics , we describe a bell type experiment involving time - delayed two - particle correlation measurements .
the indistinguishability of quantum particles results in a specific interference between different trajectories ; the non - locality in the time - delayed correlations manifests itself in the violation of a bell inequality , with the degree of violation related to the accuracy of the measurement .
in addition , we demonstrate how the interrelation between the orbital- and the spin - exchange symmetry can by exploited to infer knowledge on spin entanglement from a measurement of orbital entanglement . |
beauty of the ising model ( i m ) @xcite manifests in its simplicity .
the system considered is a network of @xmath11 interacting spins @xmath12 which energy is @xmath13 where @xmath14 is an exchange integral .
we assume homogeneous short range spin interactions , i.e. the summation in eq .
goes over pairs @xmath15 of nearest neighbours
. positive sign of @xmath16 gives ferromagnetic interaction among spins .
the minimisation of energy for temperature @xmath17 produces spin dynamics which may be described by deterministic cellular automaton with rule @xmath18 where @xmath19 denotes discrete time and summation goes over nearest neighbours of @xmath20-th spin . for finite temperature @xmath21 the deterministic rule
is replaced by a probabilistic cellular automaton with spin update rule @xmath22 described by glauber @xcite or metropolis @xcite dynamics . then the phase transition may be observed : below critical temperature @xmath23 spontaneous magnetisation @xmath24 is observed while @xmath25 for @xmath26 .
the i m was already investigated * in many ways , including monte carlo simulation @xcite , series expansion @xcite , mean - field approach @xcite or partition function technique @xcite , * and for many systems , for example : antiferromagnets @xcite , frustrated @xcite , disordered @xcite or diluted @xcite networks on complex @xcite or shuffled lattices @xcite , etc . @xcite . in this paper
the critical temperature @xmath27 is estimated for five two - dimensional lattices , basing on @xmath28 dependence , where @xmath29 denotes the time average .
the archimedean lattices are vertex transitive graphs that can be embedded in a plane such that every face is a regular polygon .
kepler showed that there is exactly eleven such graphs @xcite .
the names of the lattices are given according to the sizes of faces incident to a given vertex .
the face sizes are listed in order , starting with a face such that the list is the smallest possible in the lexicographical order . in this way
, the square lattice gets the name @xmath30 , abbreviated to @xmath31 , honeycomb is called @xmath32 and kagom is @xmath33 .
some results concerning i m on al were already presented in refs .
@xcite , however in a literature known to us the curie temperatures of several al are still missing .
critical properties of these lattices were investigated in terms of site percolation @xcite in ref .
topologies of all eleven al are given there as well .
on temperature @xmath34 expressed in @xmath5 $ ] units for @xmath10 , @xmath9 , @xmath7 , @xmath8 and @xmath31 al .
the simulations are carried out for @xmath35 spins during @xmath36 [ mcs ] .
the magnetisation @xmath37 is averaged over the last @xmath38 [ mcs].,scaledwidth=45.0% ] .al and associated critical temperatures @xmath27 . [
cols="<,<,<,<",options="header " , ] we evaluate the curie temperature @xmath27 basing on the termal dependence of magnetisation @xmath37 .
investigated systems contain about @xmath35 spins , which decorate nodes of @xmath10 , @xmath9 , @xmath7 , @xmath8 al .
the glauber dynamics @xcite is applied and the simulation takes @xmath36 monte carlo steps ( mcs ) .
one mcs is completed when all @xmath11 spins are investigated ( spin - by - spin in a type - writer order ) , wheter they should flip or not .
the time average is performed over the last @xmath38 mcs for an evaluation of @xmath37 .
the results are presented in fig .
[ fig - mvst ] .
temperature for which spontaneous magnetisation @xmath37 vanishes is accepted to be an estimation of @xmath27 .
these estimations are shown in tab .
[ tab - tc ] .
in this paper the curie temperatures for i m on all al are collected . among them , @xmath27 for @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 al are evaluated for the first time with the monte carlo simulation .
for all investigated al , the shape of @xmath39 curve ( see fig .
[ fig - norm ] ) is roughly the same as for the square lattice . in the latter case ,
an analytical expression @xcite is known @xmath40{\dfrac{\cosh^2(2/\kappa)}{\sinh^4(2/\kappa)}[\sinh^2(2/\kappa)-1]},\ ] ] where @xmath41 .
in contrast to galam mauger @xcite semi - exact formula for @xmath27 dependence on system dimensionality @xmath43 and lattice coordination number @xmath44 , we show that critical temperature for i m differ slightly for several al ( where @xmath45 ) with the same values of @xmath44 . similarly to the percolation phenomena @xcite , also for i m the dimensionality @xmath43 and the coordination number @xmath44 are not sufficient @xcite for determining the critical point @xmath27 .
k.m . is grateful to krzysztof kuakowski for many valuable and fruitful discussions .
calculations were carried out in ack - cyfronet - agh .
the machine time on hp integrity superdome is financed by the polish ministry of science and information technology under grant no .
mnii / hp_i_sd / agh/047/2004 .
heermann , _
computer simulation methods in theoretical physics _ , ( springer verlag , berlin 1990 ) ; d. stauffer , f.w .
hehl , n. ito , v. winkelmann , j.g .
zabolitzky , _ computer simulation and computer algebra _ ,
( springer verlag , heidelberg 1993 ) .
j. garca , j.a .
gonzalo , physica a * 326 * , 464 ( 2003 ) ; f. merz , j.t .
chalker , phys .
b * 65 * , 054425 ( 2002 ) ; n. ito , k. hukushima , k. ogawa , y. ozeki , j. phys . soc .
japan * 69 * , 1931 ( 2000 ) ; h .- o .
heuer , _ annual reviews of computational physics iv _ , ed .
d. stauffer , ( world scientific , singapore 1996 ) , p. 267 ; h. rieger , _ annual reviews of computational physics ii _ , ed .
d. stauffer , ( world scientific , singapore 1995 ) , p. 295 ; m. acharyya , b.k .
chakrabarti , _ annual reviews of computational physics i _ , ed .
d. stauffer , ( world scientific , singapore 1995 ) , p. 107 .
p. butera , m. comi , phys .
b * 65 * , 144431 ( 2002 ) ; p. butera , m. comi , phys . rev .
b * 56 * , 8212 ( 1997 ) ; a.j .
guttmann , i.g .
enting , j. phys .
a * 26 * , 806 ( 1993 ) ; j. oitman , c.j .
hamer , w. zheng , j. phys .
a * 24 * , 2863 ( 1991 ) .
hartmann , cond - mat/0410583 ; k. malarz , m.s .
magdo - maksymowicz , a.z .
maksymowicz , b. kawecka - magiera , k. kuakowski , int . j. mod
c * 14 * , 689 ( 2003 ) ; a.f .
ferdinand , m.e .
fisher , phys . rev . * 185 * , 832 ( 1969 ) .
j. richter , j. schulenburg , a. honecker , d. schmalfu , phys .
b * 70 * , 174454 ( 2004 ) ; s.j .
penney , v.k .
cumyn , d.d .
betts , physica a * 330 * , 507 ( 2003 ) ; p. tomczak , j. schulenburg , j. richter , a.r .
ferchmin , j. phys . : condens .
matter * 13 * , 3851 ( 2001 ) ; b. kawecka - magiera , k. kuakowski , a.z .
maksymowicz , j. magn .
mater . * 221 * , 391 ( 2000 ) ; p. tomczak , j. richter , phys .
b * 59 * , 107 ( 1999 ) .
lin , w.n .
huang , aus .
* 38 * , 227 ( 1985 ) ; k.y .
lin , j.m .
fang , phys .
a * 109 * , 121 ( 1985 ) ; k.y .
lin , t.l .
chen , chin .
* 25 * , 178 ( 1987 ) ; k.y .
lin , c.h .
kao , t.l .
chen , phys .
a * 121 * , 443 ( 1987 ) ; k.y .
lin , j.l .
chen , j. phys .
a * 20 * 5695 ( 1987 ) ; k.y .
lin , j. stat .
* 49 * , 269 ( 1987 ) ; k.y .
lin , s.c .
wang , phys .
letters a * 128 * , 143 ( 1988 ) ; k.y .
lin , phys .
letters a * 128 * , 35 ( 1988 ) ; k.y .
lin , chin . j.
* 26 * , 150 ( 1988 ) ; k.y .
lin , int .
b * 3 * , 1237 ( 1989 ) ; k.y .
lin , j. phys .
a * 22 * , 3435 ( 1989 ) .
s. c. van der marck , phys .
e * 55 * , 1228 ( 1997 ) ; s. c. van der marck , phys .
e * 55 * , 1514 ( 1997 ) ; s. c. van der marck , phys .
e * 55 * , 3732(e ) ( 1997 ) ; s. c. van der marck , int .
c * 9 * , 529 ( 1998 ) ; f. babalievski , phys .
e * 59 * , 1278 ( 1999 ) . | critical temperatures for the ferro - paramagnetic transition in the ising model are evaluated for five archimedean lattices , basing on monte carlo simulations .
the obtained curie temperatures are @xmath0 , @xmath1 , @xmath2 , @xmath3 and @xmath4 @xmath5 $ ] for @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 lattices , respectively . |
massive binary stars are invoked as the progenitors of a wide variety of astrophysical phenomena , from short and long gamma - ray bursts @xcite , type ib / c and blue type ii supernovae @xcite , to the fastest runaway o / b stars @xcite and the entire menagerie of binary systems with compact remnants : x - ray binaries , millisecond pulsar systems , and double neutron star systems ( see @xcite for a review ) .
whether binaries are necessary for the formation of all such objects is a matter of debate .
for instance , although many authors propose that type ib / c supernovae and short- and long - duration gamma - ray bursts are produced in binaries , it is still argued that their progenitors may be single stars @xcite .
observational studies can contribute to this debate by measuring the frequency , separations , and mass ratios of close massive binary systems .
although debates on the role of binaries continue , it is generally agreed that most of the progenitors of x - ray binaries and double neutron star binaries in the galactic disk must arise from massive binary systems
. population synthesis models of these systems , however , adopt very different properties for the progenitors stellar components .
these studies begin with four assumptions regarding the binary characteristics : the distribution of primary masses , the distribution of secondary masses , the distribution of orbital separations , and the distribution of eccentricities .
population synthesis studies all choose some analytic form for these distributions , generally basing them upon observational constraints on the initial and final binary system parameters
. one significant source of disagreement in these initial distributions is the choice of the secondary star mass function .
typically , the primary star mass distribution is chosen using a field initial mass function ( imf ) ( e.g. , * ? ? ?
* ; * ? ? ?
the secondary mass can be chosen through the same imf or , more commonly , from a distribution of the secondary - to - primary mass ratios : @xmath17 , where @xmath18 . in this latter formalism ,
the distribution is chosen to be proportional to the mass ratio to some power : @xmath19 between the limits @xmath20 .
note that the power - law slope , @xmath1 , is not equivalent to the power - law slope , @xmath21 , of the secondary star mass distribution , commonly expressed @xmath22 .
the relationship between @xmath17 and @xmath23 is discussed by @xcite , and it is , in general , non - trivial .
population studies disagree on the choice of @xmath1 , primarily because the choice of @xmath1 strongly effects the formation rate of key astrophysical objects .
high values maximize the formation rate of double neutron star binaries whereas low values maximize the formation rate of low - mass x - ray binaries .
generally , population synthesis studies of low - mass x - ray binaries have assumed @xmath1 values near -2.7 @xcite , whereas studies of double neutron star binaries tend to prefer a flat distribution with @xmath24 @xcite .
resolving this ambiguity is important because the choice of this value can change formation rates of , for example , neutron star binaries by nearly a factor of 100 @xcite ! this is the difference between advanced ligo detecting 10 binary inspirals per year versus 1 inspiral in 10 years .
the second critical parameter is the distribution of orbital separations , @xmath25 .
this distribution is often taken to be a power law between some inner and outer limits , @xmath4 and @xmath5 , such that @xmath26 where @xmath27 indicates a preference for large orbital separations and @xmath28 indicates a preference for small separations .
observational studies of binaries have reached widely varying , even disparate , conclusions regarding these key parameters ( reviewed in @xcite ) .
@xcite found @xmath29 for a sample of 67 massive spectroscopic binaries , suggesting that the mass ratio distribution is either flat or peaked toward massive companions . among b2b5 primaries ,
@xcite report a secondary star mass distribution consistent with the @xcite imf , corresponding to @xmath30 . in a spectroscopic study of solar type field stars ,
@xcite conclude @xmath31 for systems with @xmath32 , consistent with secondaries drawn from a field star imf .
adaptive optics imaging surveys for companions of b and a stars in the scorpius ob2 association @xcite , however , find the mass radio distribution to be well - characterized by @xmath1=-0.4 -0.5 .
@xcite report a flat ( @xmath33 ) distribution of mass ratios among single - lined spectroscopic binaries ( sb1s ) in the solar neighborhood , while double - lined spectroscopic binaries ( sb2s ) show a peak near @xmath34 .
arguments abound over whether this @xmath35 peak is real or a selection effect arising from the fact that sb2s are only identified as such if the mass and luminosity ratios are near unity @xcite several investigations have concluded that there are two populations of secondaries leading to a bimodal distribution of mass ratios . in summarizing data from the _ index catalog of visual binary stars _ @xcite @xcite and the 6th _ catalog of the orbital elements of spectroscopic binary stars _
@xcite found that @xmath36 for systems with mass ratios @xmath37 where @xmath38 .
this is roughly equivalent to secondary masses being drawn from the field star initial mass function @xcite .
for extreme mass ratio systems ,
@xmath39 , @xcite finds a flat distribution ( i.e. , @xmath40 ) .
furthermore , the turnover point , @xmath41 , may vary inversely with primary mass , but the @xcite sample did not include o stars . using radial velocity data of solar - type f7-k stars with periods @xmath42 yrs , @xcite find a broad peak in the mass ratio distribution near @xmath43 and a sharp peak near @xmath44 . in a spectroscopic study of low - mass stars , @xcite report a peak in the mass ratio distribution near @xmath45 and a smaller peak near @xmath44 . from a spectroscopic survey of eclipsing ob binaries in the @xmath46 @xcite @xcite
conclude that massive primaries have two populations of companions : a `` twin '' population with @xmath47 comprising 45% of binaries , and a population drawn from a flat mass distribution ( @xmath48 corresponding to @xmath1=0 ) comprising 55% .
this report echos earlier findings among low - mass stars that short - period ( @xmath49 d ) binaries have a `` twin '' component comprising 10% 50% of the total population @xcite .
@xcite reassessed the @xcite study using data from the 9th _ the ninth catalogue of spectroscopic binary orbits _
@xcite and concluded that the data support an excess of @xmath50 `` twin '' systems .
@xcite use a radial velocity survey of g dwarf primaries to conclude that the distribution of orbital periods for nearby binaries is broad and approximately gaussian in @xmath51 having @xmath52 and @xmath53 with @xmath54 in days .
therefore , the distribution of orbital separations should be approximately log - normal as well , with a mean of @xmath55 @xmath56 .
this result contrasts with others that indicate the distribution of orbital separations is flat in log space , @xmath57 , corresponding to @xmath58 ( i.e. , pik s law ; * ? ? ?
* ) . @xcite and @xcite measured separations @xmath59 @xmath56 for visual binaries in sco ob2 and find a flat distribution consistent with pik s law .
@xcite report a flat distribution of separations among _ hipparchos _ common proper motion stars in the solar neighborhood .
estimates for the binary fraction , @xmath0 , among massive stars vary widely , and such estimates are always lower limits given that very close , very distant , and very low - mass companions are difficult to detect . here , we use @xmath0 to mean the fraction of all primary stars having exactly one secondary .
we do not explicitly consider triple and higher order systems , even though @xcite report that a significant fraction of massive close binaries may have multiplicity @xmath60 .
@xcite report binary fractions of at least @xmath61 , and suggest that the binary fraction may be close to 100% after corrections for incompleteness . among o stars in clusters and associations , @xcite report binary fractions of @xmath6226% 55% .
@xcite find binary fractions of at least @xmath6259%
75% for o stars in clusters and associations .
@xcite tabulate multiplicity statistics for o stars in a variety of cluster environments , and they conclude that the binary fraction varies between @xmath6214% and 80% , with denser clusters having lower binary fractions . despite a wealth of observational data on parameters of binary systems , there is little apparent agreement , leaving reasonable doubt as to whether the results are reconcilable .
much of the disagreement can be explained by observational and sample selection biases .
both spectroscopic radial velocity surveys and direct imaging surveys for companions of massive primaries become incomplete at low mass ratios below @xmath63 .
secondaries in close , extreme mass ratio systems with massive primaries are presently undetectable with either technique and are likely to remain so in the near future .
thus , estimates of @xmath0 , @xmath25 , and @xmath17 at low @xmath64 must be made by extrapolating some functional form of the distribution based on data from systems with larger @xmath64 and/or large @xmath65 .
radial velocity surveys such as @xcite preferentially detect close , short - period companions ( @xmath66 @xmath56 ) while imaging surveys such as @xcite preferentially detect more distant , long - period companions .
if the fundamental parameters of binaries are different for close versus distant binaries @xcite , then the variety of reported results might be partially understood in terms of these observational selection effects ( see fig 1 of @xcite ; also @xcite ) .
furthermore , many studies do not explicitly describe the limits , @xmath4 , @xmath5 over which the results are valid , complicating comparisons between surveys .
the statistics on companions to massive stars are especially uncertain as there have been few studies of massive binaries since the survey of @xcite . for
massive primary stars , the small number of broad spectral features , coupled with rotational broadening , make radial velocity measurements difficult , and the @xcite sample may have missed binaries below velocities of 50kms@xmath67 .
more than half of their sample consisted of double - lined spectroscopic binaries , thereby introducing a bias in favor of nearly equal mass components with large relative radial velocities . about half of the sample
consisted of evolved o stars , selected on a magnitude - limited basis from throughout the sky .
given the selection biases and lack of a common origin and age for this sample , the conclusions of @xcite may not apply to the general population of zero - age massive binaries .
similarly , the conclusions of @xcite are based upon a small number of eclipsing systems in the smc , and are , therefore , most sensitive to systems of equal mass and radius so that no firm conclusions can be drawn @xcite . in this paper , we analyze data from the radial velocity survey of 120 o and early - b stars in cygnus ob2 @xcite to help resolve these disparate conclusions regarding the binary characteristics of massive stars .
the new data more than double the sample size of the @xcite study , include massive stars down to early b spectral types , and often involve more epochs of observations .
although orbital parameters for a large number of systems are not yet available , we use the observed distribution of radial velocities , in conjunction with monte carlo models , to test several popular formulations of @xmath0 , @xmath17 , and @xmath25 from the literature @xcite . using only the raw velocities of a large sample of stars from one young ob association mitigates many of the selection and evolutionary effects responsible for conflicting results from more general binary surveys and catalogs @xcite .
we find that we are able to constrain the probable values for @xmath1 and @xmath3 to a narrow region of parameter space , subject to reasonable assumptions about the boundary conditions of the mass ratios and orbital separations
. only select combinations of @xmath1 , @xmath3 , @xmath4 , @xmath5 , and @xmath0 can reproduce the levels of velocity variation seen in the data .
some canonical parameterizations for binary characteristics are inconsistent with the data and can be ruled out .
our immediate intent is to provide improved input parameters for population synthesis models that predict the rates of energetic phenomena in the milky way and at cosmic distances .
we also expect that these results will have implications for the formation scenarios of massive stars which are currently under debate @xcite .
@xcite report on a 6year radial velocity survey of 120 early - type probable members of the cygnus ob2 association .
cygnus ob2 was chosen as a target for long term radial velocity study to determine the binary fraction and mass ratio distribution among massive binaries _ in a populous young ob association where the stars are of comparable age and stellar evolutionary effects are minimized .
_ however , even in the @xmath1323 myr old cygnus ob2 , the presence of a variety of evolved stars suggests that the star formation process was non - coeval @xcite .
spectroscopic observations from the lick observatory shane telescope , the keck observatory keck i telescope , the kitt peak national observatory wiyn telescope , and wyoming infrared observatory 2.3 m telescope ( wiro ) over the period 1999 july to 2005 october were used to measure radial velocity variations of stars earlier than spectral type b3 .
sample stars were observed on at least 3 epochs over this period .
some stars have as many as 18 epochs of data .
the mean and median number of epochs are 7.6 and 7.0 respectively .
the time sampling is irregular , as imposed by telescope schedules and weather conditions .
sampling is additionally modulated by the observing season for cygnus ( june through november ) .
the data were reduced and analyzed using cross - correlation techniques , as described by @xcite .
figure [ params ] shows the analytical two - body relationship between velocity semi - amplitude and orbital period for binary systems with a 10 @xmath14 primary .
solid lines indicate the loci of systems with equivalent mass ratios , @xmath68=1.0 , 0.1 , 0.01 .
dotted lines indicate loci of common orbital separations , @xmath69 @xmath56 . the dark gray area in the upper - right demarcates the prohibited region of parameter space where @xmath70 .
the light gray area indicates the portion of parameter space sampled in this survey .
the data are sensitive to primary velocity semi - amplitudes @xmath71 km s@xmath67 , periods shorter than @xmath72 years and orbital separations @xmath73 @xmath56 .
the data presented in @xcite show at least several dozen stars with strong radial velocity variations at the level of @xmath74 km s@xmath67 and dozens of additional velocity - variable candidates with semi - amplitudes down to the survey sensitivity of @xmath75 km s@xmath67 .
there are also @xmath76 probable double - lined spectroscopic binaries , but the individual components are poorly resolved in most of our data , so these systems are , for present purposes , treated as single - lined binaries . for this analysis , we removed from the main sample of 120 objects ( table 5 of @xcite ) six stars later than b4 .
the remaining sample used for analysis is 114 objects .
these include the stars identified as probable members by @xcite plus additional early b type members identified photometrically by @xcite and classified by @xcite .
given the significant uncertainties in the photometric distances , we have undoubtedly included a few foreground and background stars .
the distribution of photometric distances ( figure 5 of @xcite ) is nearly gaussian with only a couple of outliers , suggesting that the sample does not suffer significant contamination from field stars . including a few field objects in the 114-star sample will not measurably impact the conclusions , even if the binary characteristics of field ob stars are different than the cygnus ob2 members .
figure [ vels ] plots the log of the assigned spectroscopic masses @xcite for each primary star versus the quantity @xmath77 are the maximum and minimum observed velocity for a given star .
@xmath78 is a measure of the velocity semi - amplitude of the primary , albeit an imperfect one because the velocity curves are not generally well sampled at all phases .
we regard @xmath79 as a lower limit on the true projected velocity semi - amplitude .
filled circles in figure [ vels ] denote main sequence ob stars while triangles denote evolved stars .
the error bars show the mean uncertainty in the velocity measurements over all epochs for a given star .
a small dispersion has been added to the masses in a few heavily populated mass bins to improve clarity .
one o3if star ( mt457 in the nomenclature of @xcite ) at @xmath80 @xmath14falls outside the maximum plot range .
the typical velocity precision is @xmath81 km s@xmath67 , and it varies somewhat with source luminosity and observatory / instrument combination used .
for example , the radial velocity uncertainties obtained with the wiyn telescope + hydra spectrograph are often better than those obtained with the lick 3 m telescope + hamilton echelle spectrograph because of the higher signal to noise ratio achievable with the former .
@xcite provide a more complete description of these data and place a lower limit on the binary fraction at 30% 42% .
figure [ obsdist ] shows a histogram of the observed velocities and uncertainties .
this figure illustrates the distribution of velocity dispersions , @xmath82 , calculated from the multiple measurements of the 114 ob stars ( solid line ) along with the distribution of mean velocity uncertainties , @xmath83 ( dashed line ) .
the dotted line shows the distribution of @xmath79 . the lowest velocity bin from 0 to 5 km
s@xmath67 is sparsely populated because observational errors scatter the data into higher velocity bins .
the maximum observed semi - amplitudes fall mostly between 10 and 40 km s@xmath67 , with a significant tail toward higher velocities out to @xmath1390 km s@xmath67 .
the uncertainties lie in the characteristic range 5 15 km s@xmath67 . for comparison
, figure [ garmany ] shows the distribution of properties for o - type systems from @xcite .
the left panel shows a histogram of the velocity semi - amplitudes for the new observations of o stars presented in table 3 and figure 1 of @xcite .
note the differences compared to our figure [ obsdist ] .
most striking in the garmany et al .
sample is the relative lack of stars in the velocity range 2080 km s@xmath67 and the relative excess of stars in the range 80120 km s@xmath67 .
significant selection affects may drive these differences .
the @xcite sample was magnitude limited and contained stars at a variety of evolutionary stages , including a large fraction of supergiants .
for such systems , mass exchange and transfer of angular momentum between orbital and rotational components may have altered the original binary mass ratios and separations .
therefore , the @xcite sample may not be representative of the general population of massive binaries at zero age .
the middle panel of figure [ garmany ] is a histogram of mass ratios , @xmath64 , for the o type binaries from table 3 of @xcite .
the histogram shows a preference for mass ratios near unity and is consistent with a flat ( @xmath24 ) or rising distribution ( @xmath84 ) with increasing mass ratio .
the right panel of figure [ garmany ] shows a derived semi - major axis distribution for the o type binary systems from table 3 of @xcite assuming an average orbital inclination of 60 . the dashed line is a best fit gaussian curve which peaks near @xmath85 or @xmath86 @xmath56 .
this panel shows that the o - type binaries from garmany et al .
are preferentially those with small orbital separations , as might be expected since these are the easiest to detect .
@xcite did consider whether the lack of systems with small velocity amplitudes implied a true absence of such systems or a merely limitation on their detection .
they concluded that such low - amplitude systems should have been detected as part of their survey and that the preference for equal mass components was a real effect .
we wrote a monte carlo code to simulate the radial velocity variations expected for cyg ob2 primaries based upon analytic prescriptions of binary mass ratios and orbital separations .
we initially assumed a binary fraction of @xmath9 so that every primary star has one companion .
statistically , this means that single stars are treated as systems with very low mass companions , very large separations , or very low inclinations .
however , this is a useful approach to obtain an initial lower limit on @xmath1 . for each of the 114 primary stars @xcite of mass @xmath87 and observed velocity variations characterized by @xmath79 and @xmath82 , the monte - carlo code
randomly assign a secondary star of mass @xmath88 based upon a distribution of mass ratios , @xmath17 where @xmath89 .
the mass ratios are drawn from a population described by @xmath90 where @xmath1 is the power - law index of the mass ratio distribution over the range @xmath91 .
initially , we adopt @xmath92 so that the lowest mass secondaries for our least - massive @xmath138 @xmath14 primaries are 0.16 @xmath14 and @xmath131 @xmath14 for the handful of very massive 50 @xmath14 stars . such masses are sufficiently low that the secondaries produce negligible radial reflex motion in their primaries
well below the velocity sensitivity of the data .
we consider values of @xmath1 between -3.0 and 2.0 .
@xmath84 describes a secondary star mass distribution peaked toward the primary mass .
@xmath24 corresponds to a flat distribution of secondary masses .
the monte - carlo code assigns an orbital separation , @xmath65 , drawn from a distribution described by @xmath93 \propto [ \log(r / r_{in})]^\beta$ ] with @xmath94 and @xmath65 in a.u .
pik s law ( @xmath10 ) corresponds to a parent population of orbital separations uniformly distributed in @xmath95 between @xmath96 and @xmath97 .
@xmath98 corresponds to distributions peaked toward large / small separations , respectively . because the power - law slope , @xmath3 , that best describes the orbital separations of the data will depend upon the chosen boundaries , @xmath4 and @xmath5
, we explore a small 3x3 grid of values : @xmath990.020 , 0.063 , & 0.200 @xmath56 and @xmath100100 , 1000 , & 10000 @xmath56 .
these inner limits correspond to 4.2 , 13 , and 43 @xmath101 . for a main - sequence b0 star , representative of our targets ,
these inner limits correspond to 0.56 , 1.7 , and 5.8 @xmath102 .
thereby , the smallest two values for @xmath4 are firm lower limits , representing separations where significant mass transfer and common envelope evolution would take place .
although these values are physically implausible in general , we include them to illustrate the sensitivity of the simulations to this parameter .
the third value , 0.20 @xmath56 , is the largest reasonable choice for @xmath4 , as it is larger than several of the orbital separations for short - period cyg ob2 binaries reported in @xcite and @xcite ( @xmath103 ) .
the minimum adopted outer limit , @xmath104 @xmath56 is small compared to observed orbital separations in wide binaries which may extend to several thousand @xmath56 ( e.g. , * ? ? ?
* ; * ? ? ?
the maximum adopted outer limit , @xmath105 @xmath56 , is sufficiently large that few systems are likely to remain bound at such distances , especially in the dense stellar environments where ob stars are born .
@xcite report that among nearby @xmath106 binaries , the pik s law distribution of separations fits the data out to a maximum of @xmath107 @xmath56 , beyond which the number of companions drops , consistent with gravitational disruption . in the interest of minimizing free parameters ,
we assume that the orbits have zero eccentricity .
the resulting period and velocity semi - amplitude of the simulated binary system is computed @xmath108 times , where @xmath108 is typically 100 monte - carlo iterations . for each simulated system
, the velocity curve is sampled at a random phase angle @xmath109 times , where @xmath110 , the actual number of observations of a given star .
the code assigns a random inclination angle , @xmath111 , for each system , where @xmath111 is generated by allowing the angular momentum vector of the simulated system to lie anywhere on a sphere .
the mean observational velocity uncertainty for each individual star , normally distributed about zero , is then added to the simulated velocity . as a test
, we also performed simulations using the actual uncertainties for each measurement of each star and we found that the results differed insignificantly from using the average uncertainty . for each monte carlo run with a given pair of @xmath1 and @xmath3 ,
the _ simulated _ distribution of @xmath79 and @xmath82 for 114 stars is compared to the _ observed _ distribution using a two - sided k - s test .
the probabilities that the _ simulated _ and _ observed _ distributions of @xmath79 and @xmath82 ( @xmath112 and @xmath113 , respectively ) are drawn from the same population are averaged over 100 monte carlo iterations and tabulated for later analysis .
we found that both velocity metrics , @xmath79 and @xmath82 , yield similar results and provide similar probabilistic constraints on the binary characteristics . while @xmath79 is a better measure of the true semi - amplitude , it is also less robust than @xmath82 against outliers or measurements with large uncertainties
. therefore , in subsequent discussion and figures we use the average probability , @xmath114 , obtained from the arithmetic mean of @xmath112 and @xmath113 .
in section 4 we will discuss the contraints imposed by this modeling appraoch .
the figures and discussion will show that there is degeneracy between @xmath1 and @xmath3in general , large @xmath1 coupled with large @xmath3 reproduce the velocity data approximately as well as small @xmath1 coupled with small @xmath3 .
additional constraints are needed .
one such constraint we consider is that pik s law ( @xmath10 ) obtains .
another is the fraction of type ib / c supernovae predicted by each model .
the massive binary population can also be constrained by observations of energetic phenomenae that arise from massive binary progenitors .
a well - studied example is the fraction of core - collapse supernovae that occur in stars that have lost their hydrogen envelopes type ib / c supernovae .
@xcite argued that most ib / c supernovae are produced in binary star systems where mass transfer removes most / all of the hydrogen envelope , unveiling a type ib / c supernova progenitor .
@xcite assumed that 37% of binaries were close binaries that would undergo some mass transfer .
the fraction of type ib / c supernovae predicted by population sysnthesis models must agree with observations _ and _ must self - consistently produce the levels of velocity variation seen in massive primaries , such as those in cyg ob2 . the fraction of type ib / c supernovae relative to all core - collapse supernovae is @xmath115 .
@xcite and @xcite give the observed fraction of type ib / c supernovae in sbc / d spiral galaxies such as the milky way as 15% .
the more recent works of @xcite and @xcite revise this estimate upward to 30%@xmath11611% . using binary population synthesis calculations , we can apply these fractions to further constrain the parameter space in our study .
any part of parameter space that produces a fraction of type ib / c supernovae inconsistent with the observed fraction can be ruled out .
we used the binary synthesis code developed by @xcite to calculate the total core - collapse supernova rate and the number of type ib / c ( stars that have lost their hydrogen envelopes at collapse ) supernovae .
like podsiadlowski ( 1992 ) , we find that the primary formation path for type ib / c supernovae begins with a close binary where the more massive primary envelops its companion in a common envelope phase .
the companion spirals into the envelope of the massive primary , ejecting the hydrogen envelope . after the common envelope phase
, the primary will continue to lose mass in a wind . for this paper
, we will assume any star that loses its entire hydrogen envelope has the potential to become a type ib / c supernova . because mass is lost in winds as well , even systems ( especially those at high metallicity ) that merge in the common envelope phase without ejecting
the hydrogen envelope can make type ib / c supernovae .
systems with more extreme mass ratios ( i.e. , @xmath117 ) are more likely to go through a common envelope phase and systems with such low - mass companions dominate the formation rate of type ib / c supernovae . figure [ primary ] shows the range of primary masses and orbital separations that form ib / c supernovae for 4 separate values of the system mass ratio . as @xmath64 increases , the system tends to undergo more stable mass transfer without a common envelope phase . most of these systems do not become type ib / c supernovae .
these simulations are based on our standard wind assumptions in which wind mass loss rates are set to 10% of the @xcite values . beyond 32m@xmath118
, these primary star winds will also produce type ib / c progenitors , regardless of companion mass .
however , this corresponds to only a small fraction of our type ib / c supernova rate . the other requirement for a star to produce a type ib /
c supernova is that it actually explode as a supernova .
we assume any primary with an initial mass above @xmath119 will produce a supernova .
although the total rate of type ib / c supernova does depend on this choice , the dependence on the type ii rate is nearly identical and varying this value in the @xmath120 range leads to variations in the rate of 2 - 3% for the bulk of our parameter study .
we also assume that stars that have helium cores in excess of @xmath121 collapse directly to black holes and do nt form supernovae .
varying this value of the helium core mass over the @xmath122 range yields variations in the rate of less than 1% .
this is not surprising because the number of such high - mass systems is quite small . in this paper
, we focus our binary population synthesis studies on the mass distributions of the stellar components and the orbital separation .
these distributions are varied according to range of parameters studied in the analysis of our observations and constitute over 10,000 different population synthesis calculations each including 100,000 binaries for each calculation . in all of these calculations
, we assume the eccentricity is zero .
if we instead assume the eccentricity follows a flat distribution between 0 and 0.8 , we can increase the rate by 5% .
we have already discussed the dependence of our ratios on some of the many free parameters in population synthesis : e.g. critical mass for supernova formation .
let s briefly discuss some of the remaining free parameters .
we assume a single maxwellian kick distribution with a mean value of 300kms@xmath67 . because nearly all of the type ib / c supernovae are formed in hydrogen envelope stripping in the common envelope inspiral of the companion star into the massive primary , the rate does not depend strongly on the choice of this kick .
we have varied the value of this mean value from 100kms@xmath67 to 600kms@xmath67 as well as using the bimodal kick distribution of fryer et al .
( 1998 ) and the rate does not change by more than 5% . for mass transfer ,
we use the same formalism as fryer et al .
( 1999 ) with mass transfer parameters @xmath123 and @xmath124 and a common envelope efficiency @xmath125 .
varying these values for a wide range of values ( e.g. @xmath126 ) yields variations in the ratio of 10 - 30% .
the final two uncertainties studied in our calculations are our assumed stellar radii and stellar mass loss rates . here again , we use the standard fryer et al .
( 1998 ) values .
but if we allow a factor of 2 variation ( both increase and decrease ) in these values , the ratio of type ib / c supernovae to type ii supernovae can vary by nearly 50uncertainty is one of the dominant uncertainties in the population synthesis of binary systems : the error in the stellar radius .
if the stellar radii are smaller , we produce fewer common envelope systems with a larger fraction of the remaining common envelope systems leading to mergers prior to the ejection of the hydrogen envelope .
this error dominates our 50% error .
clearly we can vary the population synthesis parameters to make many of the models fit to the data .
but if we assume we know the stellar radii and take standard values for common envelope efficiencies , our calculation errors can be limited to 510% effects , allowing us to place strong constraints on the assumed distributions of initial masses and separations .
panels within figure [ 9panela ] show probability contours ( solid lines ) depicting the likelihood , @xmath114 , that a given combination of @xmath1 and @xmath3 reproduces the distribution of velocities in the data for a minimum mass ratio of @xmath127 and binary fraction of @xmath9 .
the 3x3 grid corresponds to @xmath128 ( left column ) , @xmath129 ( middle column ) , @xmath130 ( right column ) , and @xmath131 ( upper row ) , @xmath132 ( middle row ) , @xmath133 ( lower row ) . the horizontal line in each panel marks the nominal @xmath10 ( pik s law ) distribution of separations .
the dotted contours in each panel depict the predicted fraction of type ib / c supernovae in our monte carlo models .
these contours run from 10% to 40% in increments of 5% .
the panels in figure [ 9panela ] demonstrate the sensitivity of the best - fitting solutions for @xmath1 and @xmath3 to the adopted @xmath4 and @xmath5 .
each panel contains a crescent - shaped ridge of highest likelihood contours ( @xmath134 ) .
values outside of this region drop to probabilities @xmath135% over much of the plotted parameter space . these contours
show that the best - fitting @xmath1 and @xmath3 are correlated larger orbital separations are required ( i.e. , larger @xmath3 ) when mass ratios are peaked toward unity , while smaller orbital separations are required for mass ratio distributions with secondaries drawn from a field star imf .
the ridge of peak likelihood separates parameter space into two regimes : to the left of the ridge of solid contours the models fail to produce enough high - velocity systems and are inconsistent with the data ; to the right of the peak contour ridge , the models produce too many high - velocity systems relative to the data . for a flat distribution of orbital separations described by pik s law ,
the best - fitting values for @xmath1 lie in the broad range @xmath136 depending on the choice of radial limits .
for the most reasonable choices @xmath990.0630.200
@xmath56 and @xmath132 , @xmath1 lies in the range -0.6 0.0 , broadly consistent with observational estimates of @xcite .
the dashed contours depicting @xmath1370.250.35 intersect the @xmath10 line and the solid contour ridge , indicating a ib / c fraction about twice the older observational estimates of 15% @xcite but consistent with the most recent determinations of 30%@xmath11611% @xcite .
ib / c fractions of @xmath1380.20 are inconsistent with the data in all panels except the upper - right which shows the ( perhaps unrealistically ) large radial limits of @xmath139 and @xmath140 .
it is noteworthy that , in all of the panels , mass ratio distributions described by @xmath141 are unlikely ( contours levels @xmath14230% ) even if the distribution of orbital separations favors close companions ( small @xmath3 ) .
furthermore , any @xmath143 for the most plausible radial limits in the middle or upper middle panels would require @xmath144 and @xmath145 , in disagreement with observational results .
figure [ hist9p ] shows the distribution of observed and simulated velocity semi - amplitudes , @xmath79 , ( solid and dashed lines , respectively ) for a subset of models that produce @xmath146 from the contour ridge in the middle - right panel of figure [ 9panela ] .
the left panel of figure [ hist9p ] shows a simulation with @xmath147 , @xmath148 , and the right panel shows @xmath149 , @xmath150 .
each panel is labeled with the probability that the two histograms are drawn from the same parent population .
both combinations of @xmath1 , @xmath3 yield probabilities of @xmath151 and lie along the ridge line of peak likelihood .
there is good agreement between the data and simulations based on the two - sided k - s `` d '' statistic which quantifies the maximum difference between the cumulative distribution functions of the two data sets .
an analysis of the cumulative distribution functions shows that , for these best - fitting models , the maximum differences occur in the range 1040 km s@xmath67 and 80100 km s@xmath67 in roughly equal proportion .
this means that , in some monte carlo iterations , differences in the high - amplitude tail of the velocity distributions limit the agreement between the models and data ; in other iterations , the models fail to produce the observed velocity distribution on the low - amplitude side of the histogram .
the lack of systematic trends in the velocity at which the k - s `` d '' maximum occurs may be taken as an indication that the comparison is limited by sample size and measurement uncertainties .
if , for example , the k - s `` d '' maximum difference always occurred at small @xmath79 amplitudes , we would suspect a systematic problem with the adopted velocity uncertainties which are responsible for the objects in the lowest velocity bins .
if , on the other hand , the k - s `` d '' maximum difference always occurred at large @xmath79 amplitudes , we might suspect the presence of a second population of close binaries which was responsible for the large - amplitude systems .
we will return to this latter possibility in a subsequent section .
figure [ 9panelb ] shows contour plot comparisons of the data with monte carlo simulations in the same manner as figure [ 9panela ] , except for a binary fraction of @xmath152 .
the overall level of agreement between the models and the data is lower in each panel by @xmath1310% when compared with the corresponding panel of figure [ 9panela ] .
the contours are qualitatively similar to those in figure [ 9panela ] in that the ridge of peak likelihood reaches @xmath153 in most panels , but it is shifted slightly toward larger @xmath1 and smaller @xmath3 .
variations between panels once again highlight the sensitivity of the best - fitting solutions to the adopted radial boundary conditions . for instance , the middle - right panel ( @xmath154 , @xmath155 ) shows a peak likelihood of @xmath156 for an pik s law distribution of @xmath65 , but only for @xmath157 , meaning that mass ratios must be peaked toward unity .
this is because , with fewer binaries , the observed levels of velocity variation can only be reproduced with larger mass ratios and/or smaller separations .
thus , the @xcite mass ratio distribution peaked toward unity can be consistent with the data , but only if the binary fraction is @xmath158 and only for somewhat large values of @xmath139 , i.e. , larger than many of the systems observed by @xcite .
for the most plausible radial limits ( @xmath159 , @xmath155 ) in the center panel , the most probable value for @xmath1 is @xmath160 for @xmath10 .
this implies @xmath1370.200.25 , at the lower end of current observational constraints @xcite .
any of the other panels for plausible radial limits ( upper - middle or middle - right panels ) require @xmath1380.20 , a value that is increasingly inconsistent with observational limits
. figure [ 9panelc ] shows contour plot comparisons of the data with monte carlo simulations in the same manner as figure [ 9panela ] for binary fractions of @xmath161 .
compared with the corresponding panels of figure [ 9panelb ] , the solid contours in each panel shows lower levels of agreement between models and data by about 10% .
this is especially true in the middle and upper - middle panels which represent the most plausible radial boundary limits ( @xmath162 , @xmath163 ) .
these panels also show that mass ratio distributions with @xmath141 are highly inconsistent with the data even if the the orbital separations are preferentially very small ( @xmath164 ) and @xmath165 @xmath56 .
thereby , the field imf distribution for secondary star masses is highly improbable given that @xmath165 would require many systems in common envelopes even for main - sequence primaries .
reasonable agreement with the data ( @xmath166 ) is only achievable with @xmath84 and with @xmath104 @xmath56 ( lower row ) . because @xmath104 @xmath56 is implausible ( no long - period companions ! ) and any of the panels for the most probable radial limits ( middle or upper - middle ) require @xmath1380.15 , at odds with current estimates @xcite , we conclude that figure [ 9panelc ] contains no acceptable portion of parameter space where a good fit between the models and data can be achieved
. a binary fraction as of @xmath161 or lower is highly improbable in cyg ob2 . in summary ,
the sequence of figures [ 9panela ] , [ 9panelb ] , [ 9panelc ] shows the degeneracies between @xmath1 and @xmath3 and the sensitivity of the best - fitting solutions to the adopted boundaries of the power - law parameterization of separations , @xmath25 .
distributions of @xmath64 favoring extreme mass ratios @xmath141 universally require very small separations ( @xmath164 ) in disagreement with observations supporting the canonical flat pik s law distribution .
for plausible inner and outer boundaries , the most probable values for @xmath1 lies in the range @xmath167
@xmath168 for @xmath9 and @xmath168
@xmath169 for @xmath152 .
binary fraction @xmath158 would imply low fractions of type ib / c supernovae , @xmath1380.20 , inconsistent with current estimates .
@xcite proposed that @xmath170 for @xmath37 where @xmath171 for b star primaries , and @xmath172 ( i.e. , flat ) for @xmath173 .
the relative numbers of systems above and below @xmath41 is not given .
we used an adapted monte carlo code to perform comparisons between the data and the predictions of the @xcite mass ratio distribution .
we adopted a two - component power - law distribution of @xmath64 with variable normalization for each component so that they yield the same @xmath17 at @xmath41 .
the component with @xmath173 has fixed @xmath174 while @xmath175 is allowed to vary freely .
this effectively means that the fraction of systems with @xmath37 varies with @xmath1 . for @xmath176 , 90% of systems
have @xmath37 ; for @xmath177 , 70% of systems have @xmath37 ; for @xmath178 , 48% of systems have @xmath37 .
we compared the velocity distribution of the data to the expectations of monte carlo simulations in the same manner as for the simple power - law parameterizations of @xmath17 and @xmath25 above .
we note that in the limiting case @xmath179 , the predictions of the @xcite mass ratio distribution and the power - law formulation discussed above are identical , within the noise of the monte carlo procedure employed .
figure [ 9panelh ] shows probability contour plots as a function of @xmath175 and @xmath3 for a range of @xmath4 and @xmath5 as in figure [ 9panela ] .
the ridge of peak likelihood exceeds 60% in some panels .
in particular , the center ( @xmath159 ) and middle - right ( @xmath154 ) panels show best agreement with the data .
both panels have @xmath155 @xmath56 .
the center panel shows that for the nominal pik s law distribution of separations , the most probable values for @xmath1 lie in the range @xmath180 , consistent with the hogeveen formulation for @xmath37 . however , the predicted fraction of type ib / c supernovae is @xmath1370.350.40 , at the upper edge of current estimates @xcite . in most other panels ,
@xmath141 is inconsistent with the data .
the middle - right panel shows that for the pik s law distribution of separations , the most probable values for @xmath1 lie in the range @xmath181 .
hence , the most probable value for @xmath1 for the hogeveen mass ratio distribution is much more sensitive to the choice of @xmath4 than for the simple power - law distribution of @xmath64 .
this is because the peak contour ridge in each panel is nearly horizontal , a consequence of the fact that the actual distribution of mass ratios over the whole range @xmath1820.02 1 varies comparatively little because the portion with @xmath173 is unchanging . under the hogeveen formulation
, the two - component power law with @xmath174 and @xmath183 essentially mimics a single power - law of intermediate slope , @xmath184 .
the predicted fractions of type ib / c supernovae in the middle and middle - right panels are @xmath1370.350.40 and @xmath1850.30 respectively .
figure [ 9panelh8 ] shows probability contour plots as a function of @xmath1 and @xmath3 for a range of @xmath4 and @xmath5 as in figure [ 9panelh ] , except for a binary fraction of @xmath152 .
the overall level of agreement between the models and the data is lower in each panel by @xmath186 when compared with the corresponding panel of figure [ 9panelh ] .
the contours of peak likelihood are shifted toward larger @xmath1 and smaller @xmath3 relative to figure [ 9panelh ] . in the center panel ,
the peak probabilities exceed 50% , and for an pik s law distribution of separations , the best fitting value of @xmath1 lies in the range @xmath187 , inconsistent with the @xcite value of @xmath188 . in the middle - right panel ,
the peak probabilities exceed 60% , but pik s law would require @xmath189 .
the predicted fraction of type ib / c supernovae is @xmath1370.20 - 0.25 , at the lower end of current estimates .
the lower - middle panel shows probabilities exceeding 60% for pik s law and @xmath190 .
however , this requires @xmath104 @xmath56 , and we argue previously that such a small outer bound on companion separations is not supported by observations . in summary ,
the @xcite mass ratios distribution and a binary fraction of @xmath152 produces slightly poorer overall agreement with the data . the best fitting models consistent with pik s law require @xmath191 and @xmath159@xmath192 and @xmath193 @xmath56 .
simulations with @xmath194 show comparatively poor agreement with the data , similar to figure [ 9panelc ] , and would require @xmath1380.15 , a value inconsistent with current measurements .
given ample evidence in the literature for a significant `` twin '' population among short - period binaries @xcite , we used a modified monte carlo code to simulate the expected velocity amplitudes from a two - component @xmath64 distribution .
the dominant component has a power - law distribution , @xmath19 and @xmath195^\beta$ ] between the limits @xmath159 @xmath56 and @xmath155 @xmath56 , with @xmath1 and @xmath3 being free parameters , as described previously .
the twin component comprises a fraction , @xmath196 , of the systems .
we explore a range of twin fractions , @xmath1970.40 , 0.25 , and 0.10 . allowing for the possibility that @xmath0 may be less than unity , we define @xmath196 to be the fraction of systems with @xmath50 _ among those that are binary systems_. motivated by observations that `` twin '' systems have small separations ,
presumably because they form within a causally connected region , we adopt a log - normal distribution with mean separation , @xmath198 , and gaussian width , @xmath199 .
we explore a range of typical separations , @xmath2001.0 , 3.9 , and 15 @xmath56 ( i.e. , @xmath2010.0 , 0.6 , and 1.2 ) .
the smallest of these mean separations , @xmath202 @xmath56 , is representative of the orbital distances required to produce ob star twins with orbital periods in the range few60 days like the samples used by @xcite to infer the presence of a twin population .
we adopt @xmath203 for the width of the twin population s orbital separation distribution .
( exploration of larger and smaller values for @xmath199 yielded poorer levels of agreement with the data and we do not show these results in the interest of brevity . )
the mass ratios for the twin component are uniformly distributed between @xmath1820.951.0 .
the lower mass limit for the power - law component described by @xmath19 is @xmath92 .
figure [ 9panelps ] shows a 3x3 grid of contour probability plots @xmath196=0.40 ( left column ) , @xmath196=0.25 ( middle column ) , @xmath196=0.10 ( right column ) , and @xmath20015 @xmath56 ( upper row ) , @xmath2003.9 @xmath56 ( middle row ) , @xmath2001.0 @xmath56 ( lower row ) .
the horizontal line in each panel denotes pik s law ( @xmath10 ) .
the panels in the right column having @xmath204 closely resembles the center panel of figure [ 9panela ] where no twin component is included , indicating that twin fractions of @xmath14210% do not measurably alter the results . in general
, the lower row with @xmath2001.0 @xmath56 shows relatively poor agreement with the data small mean separations yield too many large - amplitude systems .
the peak probabilities exceed 60% in several panels , most significantly in the middle panel corresponding to @xmath2003.9 @xmath56 , @xmath196=0.25 and in the middle - left panel corresponding to @xmath2003.9 @xmath56 , @xmath196=0.40 . in these panels
pik s law specifies best - fitting values for @xmath1 in the range @xmath205 .
the predicted fractions of type ib / c supernovae are large in both cases , exceeding 0.40 . in the middle - left panel showing @xmath2003.9 @xmath56 , @xmath206
, there is a broad region where the peak probabilities exceed 60% and reach 70% . in this case , if pik s law obtains , then @xmath141 , consistent with the expectations of randomly pairing secondaries from the field imf with massive primaries .
figure [ 9panelps ] illustrates that two - component @xmath64 prescription of @xcite ( 45% of systems with @xmath207 , the remainder with @xmath24 ) is not consistent with the data unless @xmath208 .
that is , the probability of @xmath24 , @xmath10 is @xmath209 for all of the panels .
the presence of any significant twin component appears to require @xmath210 .
figure [ 9panelps8 ] shows a 3x3 grid of contour probability plots similar to figure [ 9panelps ] , but for a binary fraction of @xmath152 .
the peak contours once again exceed 60% , primarily in the upper two rows of panels .
the ridge of peak likelihood is shirted toward larger @xmath1 and smaller @xmath3 relative to figure [ 9panelps ] . in most panels ,
the region with @xmath211 is smaller than in the corresponding panels of figure [ 9panelps ] , indicating slightly poorer overall agreement with the data . for an pik s law distribution of separations ,
the most probable values for @xmath1 lie in the range -1.5 0.0 for twin fractions @xmath1970.250.40 and @xmath2003.915 @xmath56 .
models with @xmath2001 @xmath56 produce generally poorer agreement with the data , having peak likelihood @xmath212 .
the two - component @xmath64 prescription of @xcite ( 45% of systems with @xmath207 , the remainder with @xmath24 ) is not consistent with the data unless @xmath213 @xmath56 . at such large distances ,
ob star twins would have much longer periods ( @xmath214 yr ) than the @xmath1330-day systems used to infer the presence of a twin population @xcite . for @xmath10 and in panels with
@xmath215 ( middle , middle - left and upper - left ) , the predicted fractions of type ib / c supernovae are @xmath1370.200.30 , in broad agreement with observations @xcite .
figure [ 9panelps6 ] shows a 3x3 grid of contour probability plots similar to figure [ 9panelps ] , but for a binary fraction of @xmath161 .
the ridge of peak likelihood is now shifted ever farther toward large @xmath1 , and the peak probabilities do not much exceed 50% in most panels , indicating poorer agreement with the data compared to figure [ 9panelps8 ] . in those panels ( middle , middle - left , upper - middle ) , pik
s law would require @xmath84 , in disagreement with the @xcite and field - star imf distribution .
furthermore , the type ib / c supernova fraction is @xmath1380.20 , in disagreement with observations .
binary fractions @xmath194 appear improbable .
power - law parameterizations of the stellar initial mass function are appropriate only over a selected mass range .
more realistic descriptions of the imf include multiple power - law components with slopes that flatten toward lower masses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we used a modified monte carlo code to simulate the velocity distribution expected if the secondary masses were drawn from a distribution described by @xcite as parameterized in equation 5 of @xcite .
figure [ 9panelscalo ] shows a 3x3 grid of contour probability plots similar to figure [ 9panela ] except that the ordinate in each panel is @xmath216 , the lower mass limit at which the @xcite imf is truncated .
the columns and rows of panels show a range of @xmath4 and @xmath5 , labeled in each panel , as in figure [ 9panela ] .
the peak probabilities marginally exceed 60% in four panels : middle - left , center , upper - middle , and upper - right .
type ib / c supernova fractions @xmath1370.20 0.40 are broadly consistent with the data . however , in all four panels , the pik s law distribution of separations shown by the solid horizontal line requires a lower mass cutoff @xmath217 .
the probabilities that the simulated data agree with the observations drop to below 40% at @xmath218 for @xmath28 .
the probabilities at low @xmath216 are somewhat higher in the left column ( @xmath219 @xmath56 ) , but we argue previously that such small inner limits , coupled with such small @xmath3 are highly improbable physically .
binary fractions @xmath220 would only exacerbate the problem , requiring even larger lower mass cutoffs to the @xcite imf . we conclude that the data are grossly inconsistent with a @xcite type imf extending below 24 @xmath14 for any plausible distribution of separations .
we tested whether a @xcite type imf might be made more consistent with the data by the addition of a `` twin '' component in the same manner described previously .
figure [ 9panelscalo25 ] shows a 3x3 grid of contour probability plots similar to figure [ 9panelscalo ] except that 25% of the stars are now `` twins '' with @xmath1820.951.0 with mean separations @xmath221 and a log - normal distribution having width @xmath203 .
the highest contours peak in excess of 60% , but these regions lie above lower mass cutoffs of @xmath216=0.8 @xmath14 for plausible values of @xmath222 .
the pik s law distribution of separations would require @xmath223 @xmath14 where it intersects the peak contour ridge .
even if 25% of the companions are twins , a secondary star mass function similar to the @xcite imf must have a lower mass cutoff at relatively large masses in order to produce the velocity distribution in the data . the implied type ib / c supernova fractions are @xmath2240.35 for the most plausible radial boundary limits ( middle and upper - middle panels ) , at the upper end of currently measured ranges @xcite .
we tested the @xcite type imf against the data with a `` twin '' fraction of @xmath206 , similar to the prescription of @xcite .
figure [ 9panelscalo40 ] shows a 3x3 grid of contour probability plots similar to figure [ 9panelscalo25 ] for this series of monte carlo comparisons .
the highest contours exceed 60% over a wide area and approach 70% . in nearly all panels ,
they extend down to cutoff masses of @xmath225 @xmath14 .
in particular , the four upper - right panels having @xmath226 and @xmath227 show that the pik s law distribution of separations intersects the ridge of peak contours in the range 0.1 0.3 @xmath14 , i.e. , in the range where the @xcite type imf is flat , or turns over for the lowest mass stars .
the implied type ib / c supernova fractions are @xmath2240.35 for these panels , at the upper end of , but consistent with , the currently measured values @xcite . in order to compare the observations to the modeled velocity distributions
we have made a number of simplifying assumptions .
some of these assumptions may account for the result that the best models agree with the data at only the 6070% level .
nevertheless , we contend that none of these assumptions affect the major conclusions .
we assume that all orbits are circular .
although this is unlikely to be the case for the majority of systems , it provides a reasonable way to estimate the mean properties of an ensemble of systems with a minimum of free parameters , especially in the absence of well - sampled velocity curves that would be required to measure eccentricities .
it is also a reasonable approximation for close , massive binaries where the orbits are likely to have circularized .
early - type systems with periods less than several days are observed to have eccentricities near zero @xcite as are systems with large fractional radii , @xmath228 where @xmath65 is the semi - major axis and @xmath102 the stellar radius @xcite .
low - mass binary systems have eccentricities near zero for periods shorter than @xmath229 days @xcite . without additional data to provide secure periods for much for our sample
, we can only say that at least @xmath214 systems appear to be in such short period orbits . furthermore , although the velocity amplitude for a primary star in an eccentric system is larger by a factor @xmath230 compared to a circular system of the same semi - major axis , the rms velocity dispersion averaged over the orbit is smaller by the same factor because the relative radial velocity variations are small over most of the orbital period .
it follows that the systematic effects of eccentric orbits cancel to zeroth order and can be neglected unless the majority of systems are highly ecentric .
we assume that each primary star has a single companion that dominates the observed primary kinematics . in reality ,
triple and quadruple systems probably exist among the sample , but they are likely to be a small fraction of all systems . among solar type stars the fraction of triple and quadruple systems is @xmath231 @xcite . the tertiary components , if any , are statistically likely to have wide orbital separations and long periods , making their observable dynamical influence minimal over few year time baselines and at @xmath1310 km s@xmath67 velocity precisions .
we assume that the measured velocity variations are due to orbital dynamics of the primary star .
stellar photospheric line profile variations may be present among some of the most massive stars , especially the evolved stars in our sample ( 31 of 114 or about 27% are post - main - sequence stars ) .
line profile variations attributed to atmospheric pulsations are observed in @xmath23277% of evolved o stars and in some be stars @xcite but rarely among dwarfs @xcite .
these phenomena could mimic the effects of bona fide orbital velocity variability in systems observed with sufficiently high spectral resolution to see these effects .
neglecting this possibility would lead to an overestimate of the binary fraction and bias the conclusions in favor a larger @xmath1 and/or smaller @xmath3 .
since our sample consists predominantly of dwarfs and non - be stars ( there are 35 be stars ) , the impact of line profile variations is minimal and is unlikely to dominate the results .
finally , we caution that the binary characteristics within a single ob association may not be representative of the binary properties in other massive starforming regions . the fraction of binaries , the distribution of mass ratios , the initial orbital separations , and their eccentricities may depend upon the _ global _ conditions in the molecular cloud from which they formed .
@xcite highlight a possible trend in the binary frequency with cluster density and richness .
for example , the dense , massive molecular clouds with densities @xmath233 @xmath234 which can produce a super star cluster with @xmath235 @xmath14 within a diameter of a few pc @xcite may produce a different population of binaries than those which generate the relatively diffuse ob associations
. this would be especially true if binaries result primarily from gravitational encounters between cloud cores and/or massive protostars early in the star formation episode .
on the other hand , it might be argued that the production of binaries is primarily a _ local _ effect driven by physics on the scale of an individual cloud core ( @xmath135 @xmath56 ) .
if massive stars form primarily through mergers of intermediate mass stars in a dense core as suggested by @xcite , then perhaps characteristics of massive binaries are set by the masses and separations of subclumps that fail to complete the merger process .
@xcite provides a review of the predictions and consequences of competing theories for massive star formation .
we have analyzed the radial velocity data spanning 6 years on a large sample of ob stars from the cygnus ob2 association to measure the probable system characteristics among massive binaries .
we have compared the radial velocity amplitudes to expectations from monte carlo models based upon several popular formulations for the mass ratios distributions in massive binaries .
these comparisons highlight the allowed parameter space and the sensitivity of the models to the key binary parameters : the true binary fraction , @xmath0 ; the power - law slope @xmath1 of the mass ratio distirbution over the range @xmath20 ; and @xmath3 , the power law slope of the orbital separations between the inner and outer radial boundaries , @xmath236 and @xmath237 .
the data improve upon prior studies by including a larger number of objects in a common association over more observational epochs .
the sample also consists of a young population with a ( mostly ) uniform age of 23 myr , thereby minimizing evolutionary and selection effects inherent in many samples .
1 . descriptions of the massive binary mass ratio distribution and orbital separation distribution derived from spectroscopic and direct imaging studies are biased by selection effects and measurement limitations .
these produce apparently conflicting prescriptions of massive binary parameters .
power - law parameterizations require explicit boundaries over which the power law is valid .
these boundaries are often insufficiently well specified in the published studies to allow meaningful comparisons given that the conclusions are sensitive to the choice of @xmath4 and @xmath5 , as shown in the figures herein .
2 . radial velocity amplitude data , such as we model here ,
are consistent with a range of mass ratio distributions and orbital radius distributions .
although the models are degenerate in @xmath1 and @xmath3 , the best fitting power - law slopes are correlated . under the assumption that the distribution of orbital separations is approximately flat ( pik s law @xmath238 ) ,
the data allow only a narrow range of @xmath1 for given inner and outer radial boundaries .
power - law parameterizations of @xmath17 and @xmath25 for plausible boundary values @xmath239 ,
@xmath240 , @xmath131 , and @xmath9 are consistent with a mass ratio distribution described by @xmath241 assuming that @xmath242 , i.e. , the flat pik s law distribution of separations .
this is consistent with @xmath243 reported for adaptive optics imaging studies of sco ob2 @xcite .
best - fitting monte carlo models are , on average , consistent with the data at the @xmath2446070% level . agreement between models and data
is limited by discrepancies between the cumulative distribution functions of the model velocity amplitudes and the data both at small velocities @xmath245 km s@xmath67 and at large velocities @xmath246 km s@xmath67 in roughly equal proportions .
this result suggests that small sample size , measurement uncertainties , and the possible presence of higher - order systems in the data all act to preclude better agreement .
power - law models with binary fractions @xmath152 fit the data slightly more poorly than models with @xmath9 , and they require larger mean mass ratios with @xmath247 . 5 .
power - law models with binary fractions @xmath194 fit the data significantly less well , having @xmath248 .
they also require extreme values of @xmath1 , @xmath3 , and @xmath249 that are inconsistent with observational studies .
the @xcite distribution of mass ratios ( @xmath188 for systems with @xmath250 and @xmath24 for systems with @xmath39 ) for @xmath129 and @xmath132 can reproduce the data at the @xmath251 level , comparable to the best - fitting power - law models with @xmath252 .
this is because the two - component @xcite @xmath64 distribution effectively mimics a power law with slope intermediate between @xmath253 and @xmath168 .
given the ad hoc nature of the this formulation , its additional parameters , and its inability to produce a superior match to the data , we find no reason to prefer it over a simpler power - law description for @xmath17 .
secondary star masses drawn from a @xcite imf are inconsistent with the data unless the imf is truncated at low masses below @xmath1324 @xmath14 .
imf truncation at such high masses is required to produce the levels of velocity variability seen in the data . _
the implication here is that the physical conditions under which massive stars form are conducive to producing relatively close binaries with correlated component masses .
mass ratio distributions involving both a twin component comprising 2540% of multiple systems and a power - law component with @xmath172 ( i.e. , flat ) for the remainder of multiple systems @xcite are inconsistent with the data unless the true binary fraction is @xmath254 .
such low binary fractions are shown to be improbable
. however , mass ratio distributions involving both a twin component comprising @xmath1340% of multiple systems and a power - law component with @xmath11 -2 -3 ( i.e. , salpeter - like ) are consistent with the data . 9 .
secondary star mass distributions comprised of two components , a power - law component and a `` twin '' component with @xmath255 , provide good agreement with the data at the level of @xmath256 , marginally better than the best - fitting single power - law models .
if the fraction of twin systems is @xmath257 and these secondaries have mean orbital separations @xmath258>=0.6\pm0.6 $ ] , then the data are consistent with the remainder of the companions being drawn from the canonical miller - scalo -type field star imf
. this formalism simultaneously predicts a type ib / c supernova fraction of @xmath1370.300.40 , consistent with the most recent estimates .
_ this two - component description may allow for the production of massive binary phenomena such as the large number of double neutron star systems that seem to be required by the data ( see * ? ? ?
* for a review ) while simultaneously permitting the production of low - mass x - ray binaries at the expected rates .
the best - fitting binary parameters also predict that the fraction of core - collapse supernovae that are type ib / c supernovae is least 30% ( assuming single stars do not produce type ib / c supernovae ) . if the actual ib / c supernova fraction were 15% , the allowed binary parameter space would require binary systems where the two components were all nearly the same mass ( high @xmath1 ) and with orbital separations well above pik s law ( @xmath259 ) .
such values , although possible , seem unlikely . more likely
is that the binary observations indirectly support the most recent supernova observations suggesting that the ib / c supernova fraction may be as high as 3040% .
observational work that better defines the distribution of binary separations , mass ratios , and binary fractions will help to reduce degeneracies in the types of model constraints used here .
thereby , such data will allow a more precise specification of initial binary parameters .
optical interferometric observations of companions to massive stars in the radial range @xmath260 @xmath56 would help verify that the pik s law distribution of separations extends to smaller orbital distances .
such data may also begin to directly identify `` twin '' systems with typical separations @xmath131few @xmath56 .
the continuing cyg ob2 radial velocity survey @xcite will also measure orbital parameters , modulo inclinations , for a sizable sample of massive binaries with a common origin , permitting tighter constraints on the binary separations and companion masses .
we thank the time allocation committees of the lick , keck , wiyn , and wiro observatories for generous observing allocations and stan woosley for making this project possible .
we acknowledge helpful conversations with peter conti , steve vogt , laura penny , bob mathieu , doug gies , and ed van den heuvel .
we acknowledge dan kiminki for his assistance with the cygnus ob2 survey data in the preparation of this manuscript .
we thank the referee , ronald webbink , for his substantial investment of time on comments that greatly improved this paper .
we are grateful for support from the national science foundation through the research experiences for undergraduates ( reu ) program grant ast-0353760 and through grant ast-0307778 . the work of c. f. was in part under the auspices of the u.s .
dept . of energy , and supported by its contract w-7405-eng-36 to los alamos national laboratory and by national science foundation under grant no .
phy99 - 07949 .
abt , h. a. , & levy , s. g. 1976 , apjs , 30 , 273 abt , h. a. , & levy , s. g. 1978 , apjs , 36 , 241 abt , h. a. , gomez , a. e. & levy , s. g. 1990 , apjs , 74 , 551 abt , h. a. 1983 , ara&a , 21 , 343 batten , a. h. , fletcher , j. m. , & maccarthy , d. g. 1989 , eight catalog of the orbital elements of spectroscopic binary systems , victoria , b.c .
belczynski , k. kalogera , v. , & bulik , t. 2002 , apj , 572 , 407 bhattacharya , d , & van den heuvel , e. p. j. 1991 , phys .
reports , 201 , 1 billett , o. h. , hunter , d. a. , elmegreen , b. g. 2002 , aj , 123 , 1454 blaauw , a. 1961 , ban , 15 , 265 bonnell , i. a. , bate , m. r. , & zinnecker , h. 1998 , mnras , 298 , 93 cappellaro , e. , turatto , m. , tsvetkov , d. yu . , bartunov , o. s. , pollas , c. , evans , r. , hamuy , m. 1997 , a&a , 322 , 431 cappellaro , e. , evans , r. , & turatto , m. , 1999 , a&a , 351 , 459 duquennoy , a. & mayor , m. 1991 , a&a , 248 , 485 eggleton , p. p. , fitchett , m. j. , & tout , c. a. 1989 , apj , 347 , 998 elmegreen , b. g. , & efremov , y. n. 1997 , apj , 480 , 235 fisher , j. , schrder , k .- p . , & smoth , r. c. 2005 , mnras , 361 , 495 fryer , c. l. , burrows , a. , & benz , w. 1998 , apj , 496 , 333 fryer , c.l . , & woosley , s.e .
1998 , apj , 502 , l9 fryer , c. l. , woosley , s. e. , & hartmann , d. h. 1999 , apj , 526 , 152 fryer , c. l. , rockefeller , g. , hungerford , aimee , & melia , f. 2006 , apj , 638 fullerton , a. w. , gies , d. r. , & bolton , c. t. 1996 , apjs , 103 , 475 garmany , c. d. , conti , p. s. , & massey , p. 1980 , apj , 242 , 1063 garcia , b. & mermilliod , j. c. 2001 , a&a , 368 , 122 gies , d. r. 1987 , apjs , 64 , 545 giuricin , g. , mardirossoan , f. , & mezzetti , m. 1984 , a&a , 134 , 365 goldberg , d. , & mazeh , t. 1994 , a&a , 282 , 801 halbwachs , j. l. 1987 , a&a , 183 , 234 halbwachs , mayor , m. , udry , s. , & arenou , f. 2003 , a&a , 397 , 159 hanson , m. m. 2003 , apj , 597 , 957 harries , t. j. , hilditch , r. w. , howarth , i. d. 2003 , mnras , 339 , 157 heger , a. , fryer , c. l. , woosley , s. e. , langer , n. , hartmann , d. h. 2003 , apj , 591 , 288 hillwig , t. c. , gies , d. r. , bagnolo , w. g. , huang , w. , mcswain , m. v. , wingert , d. w. 2006 , apj , in press hirschi , r. , meynet , g. , & maeder , a. 2005 , a&a , 443 , 581 hogeveen , s. j. 1990 , ap&ss , 173 , 315 hogeveen , s. j. 1992 , ap&ss , 196 , 299 humphreys , r. m. & mcelroy , d. b. 1984 , apj , 284 , 565 jeffers , h. m. , van den bos , w. h. , & greeby , f. m. 1963 , index catalog of visual double stars , lick observatory , mount hamilton kalogera , v. , & webbink , r. f. 1998 , apj , 493 , 351 kalogera , v. , beclzynski , k. , kim , c. , oshaughnessy , r. , & willems , b. 2007 , phys . reports , 442 , 75 kiminki , d. , kobulnicky , h. a. , kinemuchi , k. , irwin , j. s. , fryer , c. l. berrington , r. c. , uzpen , b. monson , a. j. , pierce , m. a. , & woosley , s. e. 2007 , apj , in press kobulnicky , h. a. & johnson , k. e. 1999 , apj , 527 , 154 kouwenhoven , m. b. n. , brown , a. g. a. , zinnecker , h. , kaper , l. , & portegies zwart , s. f. 2005 , a&a , 430 , 137 kouwenhoven , m. 2006 , phd thesis , university of amsterdam krumholz , m. r. 2006 , asp conf . ser . in press krumholz , m. r. , mckee , c. f. , & kleim , r. i. 2005 , apj , 618 , l33 kroupa , p. , tout , c. a. , & gilmore , g. 1991 , mnras , 251 , 293 kroupa , p. , tout , c. a. , & gilmore , g. 1993 , mnras , 262 , 545 larson , r. b. 2001 , iau symp .
200 , 93 lanz , t. , & hubent , i. 2003 , apjs , 146 , 417 lpine , s. , & bongiorno , b. 2007 , aj , 133 , 889 leaman , et al .
2007 , in preparation li , w. , wang , x. , van dyk , s. d. , cuillandre , j - c . ,
foley , r. j. , filippenko , a. v. 2007 , astro - ph/0701049 lucy , l. b. , & ricco , e. 1979 , aj , 84 , 401 lucy , l. b. 2006 , a&a , 457 , 629 martins , f. , & schaerer , d. , & hillier , d. j. 2005 , a&a , 436 , 1049 mannucci , f. , della valle , m. , panagia , n. , cappellaro , e. , cresci , g. , maiolino , r. , petrosian , a. , turratto , m. , 2005 , a&a , 433 , 807 mason , b. d , gies , d. r. , hartkope , w. i. , bagnolo , w. g. , brummelaar , t. , mcalister , h. a. 1998 , aj , 115 , 821 massey , p. , thompson , a.b . , 1991 , aj 101 , 1408 mazeh , t. , goldberg , d. , duquennoy , a. , & mayor , m. 1992 , apj , 401 , 265 meibom , s. , & mathieu , r. d. 2005 , apj , 620 , 970 miller , g. e. , & scalo , j. m. 1979 , apjs , 41 , 513 north , p. , & zahn , j .- p . 2003 , a&a , 405 , 677 pik , e. j. , 1924 , tartu obs . publ . , 25 oconnell , r. w. , gallagher , j. s. iii , hunter , d. a. 1994 , apj , 433 , 650 pan , k. , tan , h. , & shan , h. 1998 , a&a , 335 , 179 penrod , g. d. 1986 , pasp , 98 , 35 petrie , r. m. 1960 , ann .
astrophysics , 23 , 744 pfahl , e. rappaport , s. , podsiadlowski , p. 2003
, apj , 597 , 1036 pinsonneault , m. h. & stanek , k. z. 2006 , apj , 639 , l67 , podsiadlowski , ph . , joss , p. c. , hsu , j. j. l. 1992 , apj , 391 , 246 podsiadlowski , ph . , rappaport , s. , pfahl , e. d. 2002 , apj , 565 , 1107 pourbaix , d. , tokovinin , a. a. , batten , a. h. , fekel , f. c. , hartkopf , w. i. , levato , h. , morrell , n. i. , torres , g. , udry , s. 2006 , a&a , 424 , 727 poveda , a. , allen , c. , & parrao , l. 1982 , apj , 258 , 589 salpeter , e. e. 1955 , apj , 121 , 161 scalo , j. m. 1986 , fund .
phys . , 11 , 1 scarfe , c. d. 1986 , jrasc , 80 , 257 shatsky , n. , & tokovinin , a. 2002 , a&a , 382 , 92 tokovinin , a. a. , 2000 , a&a , 360 , 997 tout , c. 1991 , mnras , 250 , 701 trimble , v. 1974 , aj , 79 , 967 vogt , s. s. , & penrod , g. d. 1983 , apj , 275 , 661 wolfire , m. g. , & cassinelli , j. p. 1987 , apj , 319 , 850 woosley , s. e. , langer , n. , & weaver , t. a. 1995 , apj , 448 , 315 zahn , j .-
1977 , a&a , 57 , 383 | we constrain the properties of massive binaries by comparing radial velocity data on 114 early - type stars in the cygnus ob2 association with the expectations of monte carlo models .
our comparisons test several popular prescriptions for massive binary parameters while highlighting the sensitivity of the best - fitting solutions to the adopted boundary conditions .
we explore a range of true binary fraction , @xmath0 , a range of power - law slopes , @xmath1 , describing the distribution of companion masses between the limits @xmath2 and 1 , and a range of power - law slopes , @xmath3 , describing the distribution of orbital separations between the limits @xmath4 and @xmath5 .
we also consider distributions of secondary masses described by a miller - scalo type initial mass function ( imf ) and by a two - component imf that includes a substantial `` twin '' population with @xmath6 .
several seemingly disparate prescriptions for massive binary characteristics can be reconciled by adopting carefully chosen values for @xmath0 , @xmath4 , and @xmath5 .
we show that binary fractions @xmath7 are less probable than @xmath8 for reasonable choices of @xmath4 and @xmath5 .
thus , the true binary fraction is high . for @xmath9 and a distribution of orbital separations near the canonical pik s law distribution ( i.e.
, flat ; @xmath10 ) , the power law slope of the mass ratio distribution is @xmath11 - 0.6 0.0 . for @xmath12
, @xmath1 is somewhat larger , in the range -0.4 1.0 . in any case
, the secondary star mass function is inconsistent with a miller - scalo -like imf unless the lower end is truncated below @xmath1324 @xmath14 .
in other words , massive stars preferentially have massive companions .
the best fitting models are described by a salpeter or miller - scalo imf for 60% of secondary star masses with the other @xmath15 of secondaries
having @xmath16 , i.e. , `` twins '' .
these best - fitting model parameters simultaneously predict the fraction of type ib / c supernovae to be 3040% of all core - collapse supernovae , in agreement with recent observational estimates . |
let us start with isaac newton .
he formulated his gravity law applicable to two point particles .
it took him 20 years to extend his law for solid spheres with non - zero radii . in 1905 , albert einstein formulated his special relativity and was interested in how things look to moving observers .
he met niels bohr occasionally .
bohr was interested in the electron orbit of the hydrogen atom .
then , they could have talked about how the orbit looks to a moving observer @xcite , as illustrated in fig .
[ newton ] .
if they did , we do not know anything about it .
indeed , it is for us to settle this bohr - einstein issue .
the purpose of the present paper is to discuss whether wigner s 1939 paper on the lorentz group provides the framework to address the internal space - time symmetries of particles in the lorentz - covariant world .
this question is far more important than whether wigner deserved a nobel prize for this paper alone . for many years since 1963 , many people claimed that wigner s 1939 paper is worthless because he did not get the nobel prize for it
let us respond to this fatalistic view .
einstein did not get the prize for his formulation of special relativity in 1905 .
does this mean that einstein s special relativity worthless ?
we shall return to this question in the appendix .
however , it is quite possible that wigner started subject , but did not finish it .
if so , how did this happen ? in his 1939 paper
@xcite , wigner considered the subgroups of the lorentz group whose transformations leave the four - momentum of a given particle invariant .
these subgroups are called wigner s little groups and dictate the internal space - time symmetries in the lorentz - covariant world .
he observed first that a massive particle at rest has three rotational degree of freedom leading to the concept of spin .
thus the little group for this massive particle is like @xmath0 .
how about this massive particle moving in the @xmath1 direction .
we could settle this issue easily .
wigner observed also that a massless particle can not be brought to its rest frame , but he showed that the little group for the massless particle also has three degrees of freedom , and that this little is locally isomorphic to the group @xmath2 or the two - dimensional euclidean group .
this means that generators of this little group share the same set of closed commutation relations with that for two - dimensional euclidean group with one rotational and two translational degrees of freedom .
it is not difficult to associate the rotational degree of freedom of @xmath2 to the helicity of the massless particle .
however , what is the physics of the those two translational degrees of freedom ?
wigner did not provide the answer to this question in his 1939 paper @xcite .
indeed , this question has a stormy history , and the issue was not completely settled until 1990 @xcite , fifty one years after 1939 , or 27 years after his nobel prize in 1963 . direction .
however , the physics of the first matrix ( matrix a ) was not completely understood until 1990 , 51 years after 1939 . ]
cccc & & & + + & & & + & massive , slow & covariance & massless , fast + & & & + energy- & & einstein s & + momentum & @xmath3 & @xmath4 & @xmath5 + & & + internal & @xmath6 & & @xmath6 + space - time & & wigner s & + symmetry & @xmath7 & little group & gauge transformations + & & + + for many years , the major complaint had been that his little groups could not explain the maxwell field .
is it possible to construct electromagnetic four - potential and the maxwell tensor as representations of representations of wigner s little group for massless particles ? to answer this question ,
let us go to one of his matrices in his paper , given in fig .
[ annals ] .
it is easy to see that matrix b is for a lorentz boost along the @xmath1 direction .
matrix a leaves the four - momentum of the massless particle invariant .
what else does this matrix do ? in 1972 @xcite , kuperzstych showed that it performs a gauge transformation when applied to the electromagnetic four - potential , but he did not see this as wigner s problem .
indeed , this question was not completely answered until 1990 @xcite . in the present paper , we point out the complete understanding of this matrix leads to result given in table [ gen11 ] contained the paper i published with my younger colleagues in 1986 @xcite . as
einstein s energy - momentum leads to its expressions both in the small - momentum and large - momentum limits , wigner s little groups explain the internal space - time symmetries for the massive particle at rest as well as for the massless particle , as summarized in table [ gen11 ] . from sec .
[ little ] to sec .
[ o3e2 ] , technical details are given .
the present author gave many lectures on this subject in the past .
in this report , he explains the same subject at a public - lecture level by avoiding group theoretical words as much as possible . since this paper deals with a sensitive issue , it is appropriate to mention his background and as well as his experience in dealing with those people who did not agree with him . in sec .
[ little ] , wigner s little groups are spelled out in the language of four - by - four matrices . in sec .
[ spinhalf ] , the two - by - two representation if given for spin - half particles .
the gauge transformation is defined for this two - by - two representation . in sec .
[ fourvec ] , it is shown that the gauge transformation defined in the two - by - two representation leads to the gauge transformation applicable to the four - potential for photons . in sec .
[ o3e2 ] , the @xmath0-like little group for massive particles . in the appendix
, it is noted that this paper deals with a very serious historical issue .
the question then is whether the present author is qualified to write about this issue .
thus , it is appropriate to explain how and why he had to study wigner s paper .
a more interesting story is how much resistance he had to face in making his results known to wigner and to the world .
if we use the four - vector convention @xmath8 , the generators of rotations around and boosts along the @xmath1 axis take the form @xmath9 respectively .
we can also write the four - by - four matrices for @xmath10 and @xmath11 for the rotations around the @xmath12 and @xmath13 directions , as well as @xmath14 and @xmath15 for lorentz boosts along the @xmath12 and @xmath13 directions respectively @xcite .
these six generators satisfy the following set of commutation relations .
@xmath16 = i\epsilon_{ijk } j_{k } , \qquad \left[j_{i } , k_{j}\right ] = i\epsilon_{ijk } k_{k } , \qquad \left[k_{i } , k_{j}\right ] = -i\epsilon_{ijk } j_{k}.\ ] ] this closed set of commutation relations is called the lie algebra of the lorentz group .
the three @xmath17 operators constitute a closed subset of this lie algebra .
thus , the rotation group is a subgroup of the lorentz group .
in addition , wigner in 1939 considered a subgroup generated by @xcite @xmath18 these generators satisfy the closed set of commutation relations @xmath19 = 0 , \qquad \left[j_{3 } , n_{1}\right ] = in_{2 } , \qquad \left[j_{3 } , n_{2}\right ] = -in_{1}.\ ] ] group and the cylindrical group .
they share the same lie algebra , but only the cylindrical group leads to a geometrical interpretation of the gauge transformation . ] as wigner observed in 1939 @xcite , this set of commutation relations is just like that for the generators of the two - dimensional euclidean group with one rotation and two translation generators , as illustrated in fig .
[ cylin ] . however , the question is what aspect of the massless particle can be explained in terms of this two - dimensional geometry .
indeed , this question has a stormy history , and was not answered until 1987 . in their paper of 1987 @xcite ,
kim and wigner considered the surface of a circular cylinder as shown in fig .
[ cylin ] . for this cylinder , rotations are possible around the @xmath1 axis .
it is also possible to make translations along the @xmath1 axis as shown in fig .
[ cylin ] .
we can write these generators as @xmath20 applicable to the three - dimensional space of @xmath21 they then satisfy the closed set of commutation relations @xmath22 = 0 , \qquad \left[l_{3 } , q_{1}\right ] = iq_{2 } , \qquad \left[l_{3 } , q_{2}\right ] = -iq_{1}.\ ] ] which becomes that of eq.([eq200 ] ) when @xmath23 and @xmath24 are replaced by @xmath25 and @xmath26 of eq.([eq005 ] ) respectively . indeed , this cylindrical group is locally isomorphic to wigner s little group for massless particles .
let us go back to the generators of eq.([eq005 ] ) .
the role of @xmath26 is well known .
it is generates rotations around the momentum and corresponds to the helicity of the massless particle .
the @xmath27 and @xmath28 matrices take the form @xcite @xmath29 the transformation matrix is @xmath30 in his 1939 paper @xcite , wigner observed that this matrix leaves the four - momentum of the massless particle invariant as can be seen from @xmath31 but he never attempted to apply this matrix to the photon four - potential .
it is interesting to note that in 1976 noted that this form in applicable to the four - potential while making rotation and boosts whose combined effects do not change the four four - momentum @xcite of the photon . in 1981 ,
han and kim carried out the same calculation within the framework of wigner s little group @xcite .
kuperzstych s conclusion was that the four - by - four matrix of eq.([eq101 ] ) performs a gauge transformation when applied to the photon four - potential , and han and kim arrived at the same conclusion .
let us see how this happens .
let us next consider the electromagnetic wave propagating along the @xmath1 direction : @xmath32 and apply the @xmath33 matrix to this electromagnetic four - vector : @xmath34 which becomes @xmath35 if the four - vector satisfies the lorentz condition @xmath36 , this expression becomes @xmath37 the net effect is an addition of the same quantity to the longitudinal and time - like components while leaving the transverse components invariant .
indeed , this is a gauge transformation .
let us go back to the lie algebra of the lorentz group given in eq.([eq002 ] ) .
it was noted that there are six four - by - four matrices satisfying nine commutation relations .
it is possible to construct the same lie algebra with six two - by - two matrices @xcite .
they are @xmath38 where @xmath39 are the pauli spin matrices . while @xmath17 are hermitian
, @xmath40 are not .
they are anti - hermitian .
since the lie algebra of eq.([eq002 ] ) is hermitian invariant , we can construct the same lie algebra with @xmath41 this is the reason why the four - by - four dirac matrices can explain both the spin-1/2 particle and the anti - particle .
thus the most general form of the transformation matrix takes the form @xmath42 and this transformation matrix is applicable to the spinors @xmath43 in addition , we have to consider the transformation matrices @xmath44 applicable to @xmath45 with this understanding , let us go back to the lie algebra of eq.([eq002 ] ) . here again the rotation generators satisfy the closed set of commutation relations : @xmath46 = i\epsilon_{ijk } j_{k } , \qquad \left[\dot{j}_{i } , \dot{j}_{j}\right ] = i\epsilon_{ijk } \dot{j}_{k } .\ ] ] these operators generate the rotation - like @xmath47 group , whose physical interpretation is well known , namely the electron and positron spins . here also we can consider the @xmath2-like subgroup generated by @xmath48 the @xmath27 and @xmath28 matrices take the form @xmath49 on the other hand , in the `` dotted '' representation , @xmath50 there are therefore two different @xmath51 matrices : @xmath52 and @xmath53 these are the gauge transformation matrices applicable to massless spin-1/2 particles @xcite
. here are talking about the dirac equation for with four - component spinors .
the spinors @xmath54 and @xmath55 are gauge - invariant since @xmath56 as for @xmath57 and @xmath58 , @xmath59 & { } & \dot{d}(u , v ) \dot{\chi}_{+ } = \dot{\chi}_{+ } + ( u + iv)\dot{\chi}_{- } .\end{aligned}\ ] ] they are not invariant under the @xmath51 transformations , and they are not gauge - invariant .
thus , we can conclude that the polarization of massless neutrinos is a consequence of gauge invariance , as illustrated in fig .
[ neupol ] .
we are familiar with the way in which the spin-1 vector is constructed from the spinors in non - relativistic world .
we are now interested in constructing four - vectors from these spinors .
first of all , with four of the spinors given above , we can start with the products .
@xmath60 resulting in spin-0 scalars and four - vectors and four - by - four tensors for the spin-1 states @xcite .
the four - vector can be constructed from the combinations @xmath61 and @xmath62 among them , let us consider the combinations , let us consider the four resulting from @xmath63
. among them , as far as the rotation subgroup is concerned , @xmath64 , and @xmath65 are like @xmath66 and @xmath67 respectively , and and invariant under lorentz boosts along the @xmath1 direction .
in addition , we should consider @xmath68 which are invariant under rotations around the @xmath1 axis .
when the system boosted along the @xmath1 direction , these combinations are transformed like @xmath1 and @xmath69 directions respectively . with these aspects in mind ,
let us consider the matrix @xmath70 and write the transformation matrix @xmath71 of eq.([eq705])as @xmath72 if four matrix elements are complex numbers , there are eight independent parameters .
however , the condition @xmath73 reduces this number to six .
the lorentz group starts with six degrees of freedom .
it is then possible to write the four - vector @xmath74 as @xmath75 with its lorentz - transformation property @xmath76 the four - momentum can also be written as @xmath77 with the the transformation property same as that for @xmath78 given in eq.([eq717 ] ) . with this understanding , we can write the photon four - potential as @xmath79 let us go back the two - by - two matrices @xmath80 and @xmath81 given in eqs.([d201 ] ) and ( [ d202 ] ) .
we said there that they perform gauge transformations on massless neutrinos .
it is indeed gratifying to note that they also lead to the gauge transformation applicable to the photon four - potential .
@xmath82 this results in @xmath83 if we apply the lorentz condition @xmath84 , this matrix becomes @xmath85 this result is the same as the gauge transformation in the four - by - four representation given in eq.([eq150 ] ) .
in his 1939 paper @xcite , wigner discussed his little groups for massive and massless particles as two distinct mathematical devices .
indeed , inn and wigner in 1953 initiated of the unification of these little groups by observing considering a flat plane tangent to a sphere , while the plane and sphere correspond to the @xmath2 and @xmath0 symmetries respectively @xcite .
this unification was completed in 1990 @xcite .
the issue is whether the @xmath2-like little group can be obtained as a zero - mass limit of the @xmath0-like little group for massive particles .
another version of this limiting process is given in sec .
[ o3e2 ] of the present report . as for the internal space - time symmetry of particles ,
let us go beck to bohr and einstein .
bohr was interested in the electron orbit of the hydrogen atom while einstein was worrying about how things look to moving observers .
they met occasionally before and after 1927 to discuss physics .
did they talk about how the stationary hydrogen atom would look to a moving observer ?
it they did , we do not know about it .
this problem is not unlike the case of newtona law of gravity .
newton worked out the inverse square law for two point particles .
it took him 20 years to work out the same law for extended objects such and the sun and earth , as illustrated in fig .
[ newton ] . in 1905 , einstein formulated his special relativity for point particles .
it is for us to settle the issue of how the electron orbit of the hydrogen atom looks to moving observers .
indeed , the circle and ellipse as given in fig . [ newton ] have been used to illustrate this this relativistic effect .
however , these figures do not take into account the fact that the electron orbit had been replaced by a standing wave .
indeed , we should learn how to lorentz - boost standing waves . yes , we know how to construct standing waves for the hydrogen atom .
do we know how to lorentz - boost this atom ?
the answer is no .
however , we can replace it with the proton without changing quantum mechanics . both the hydrogen atom and the proton are quantum bound states , but
the proton can be accelerated . while the coulomb force is applicable to the hydrogen , the harmonic oscillator potential is used as the simplest binding force for the quark model @xcite .
we can switch the coulomb wave functions with oscillator wave functions without changing quantum mechanics .
this problem is illustrated in fig .
[ quapar ] .
then it is possible to construct the oscillator wave functions as a representation of wigner s little group @xcite . in this two - by - two representation , the lorentz boost along the positive direction is @xmath86 the rotation around the @xmath13 axis is @xmath87 then , the boosted rotation matrix is @xmath88 if @xmath89 becomes very large , and this matrix is to remain finite , @xmath90 has to become very small , and this expression becomes @xcite @xmath91 with @xmath92 this expression becomes @xmath93 in this two - by - two representation , the rotation around the @xmath1 axis is @xmath94 respectively .
thus @xmath95 which becomes @xmath96 with @xmath97 here , we have studied how the little group for the @xmath0-like little group the massive particle becomes the @xmath2-like little group for the massless particle in the infinite-@xmath89 limit .
what does this limit mean physically ?
the parameter @xmath89 can be derived from the speed of of the particle .
we know @xmath98 , where @xmath99 is the speed of the particle .
then @xmath100 where @xmath101 and @xmath102 are the mass and the momentum of the particle respectively . if @xmath101 is much smaller than @xmath103 , @xmath104 which becomes large when @xmath101 becomes very small .
thus , the limit of large @xmath89 means the zero - mass limit .
let us carry our the same limiting process for the four - by - four representation . from the generators of the lorentz group , it is possible to construct the four - by - four matrices for rotations around the @xmath13 axis and lorentz boosts along the @xmath1 axis as @xcite @xmath105 respectively .
the lorentz - boosted rotation matrix is @xmath106 which can be written as @xmath107 while @xmath108 , this boosted rotation matrix becomes a transformation matrix for a massless particle when @xmath89 becomes infinite . on the other hand ,
it the matrix is to be finite in this limit , the angle @xmath90 has to become small . if we let @xmath109 as given in eq.([gamma ] ) , this four - by - four matrix becomes @xmath110 this is the lorentz - boosted rotation matrix around the @xmath13 axis .
however , we can rotate this @xmath13 axis around the @xmath1 axis by @xmath111
. then the matrix becomes @xmath112 this matrix becomes @xmath80 of eq.([eq101 ] ) , if replace @xmath113 and @xmath114 with @xmath115 and @xmath99 respectively , as given in eq.([eq602 ] ) .
in this report , i am dealing with a very serious issue in physics .
the question is whether i am qualified to talk about wigner s 1939 paper .
the reader of this article is not likely to be the first one to raise this issue .
louis michel and arthur wightman were among the most respected physicists on the lorentz group , and their photos are in fig .
[ micwi ] . in 1961
, while i was a graduate student at princeton , i was in wightman s class .
i learned from him the abc of the lorentz group .
wightman gave the same set of lectures in france , and he published an article in french @xcite . in 1962 , louis michel gave a series of very comprehensive lectures at a conference held in istanbul .
indeed , i learned from his lecture note @xcite how the inhomogeneous lorentz group is different from the homogenous lorentz group .
both michel and wightman became upset when i was meeting wigner during the period from 1985 to 1990 .
wightman sent me a letter telling me my papers on wigner s 1939 paper are wrong . in particular , he said the table given in table [ gen11 ] is wrong .
i assume he told the same story to wigner because his office and wigner s office were in the same building on the campus of princeton university .
louis michel became very angry when i had to tell him i could carry out my wigner program without his permission .
he told me he did not like what i say in table [ gen11 ] .
he even wrote a letter to john s. toll telling him to reduce my position at the university of maryland in 1987 .
toll was the chancellor of the state - wide university of maryland system at that time .
he was john a. wheeler s student at princeton university and came to the university of maryland in 1953 to build the present form of the physic department . in 1962 , he hired me as an assistant professor one year after my phd degree at princeton .
toll became very happy whenever wigner came to maryland at my invitation , as indicated in fig .
[ twk ] . in spite of those hostile reactions from michel and wightman , wigner liked table [ gen11 ] , and continued listening to me .
john s. toll continued supporting my position at the university of maryland . in spite of what i said above i still like michel and wightman .
they were great teachers to me .
stephen adler and gerald feinberg were also very influential physicists during the period from 1960 to 1990 .
i knew them well . in 1981 ,
when i submitted a paper with my younger colleagues on the wigner issue to the physical review letters , feinberg wrote a referee report saying wigner s 1939 paper is a useless mathematical exposition having nothing to do with physics .
he was so proud of what he was saying that he revealed his name in his report , while the referees are anonymous .
thus , he deserves to be mentioned in the present report .
since feinberg did not give other reasons in his report , we resubmitted the paper asking the editor to examine its scientific contents . at that time , adler was in the editorial position to make the final decision .
adler said he agreed with feinberg without making any comments of his own .
in other words , adler was also saying that wigner s 1939 paper is worthless . in effect , both adler and feinberg were telling us not to waste my time because wigner did not get the nobel prize for this paper .
steven weinberg was different . in 1964 , he published a series of papers on the spin states that can be constructed from wigner s little groups @xcite .
indeed , he realized that wigner s little groups are for the internal space - time symmetries .
as for massless particles , weinberg realized the matrix a of fig .
[ annals ] was troublesome , and constructed his `` state vectors '' which are independent of this matrix @xcite .
does weinberg s result bring wigner s paper closer to the maxwell theory ? in the maxwell formalism , it is possible to construct gauge - independent states , namely electromagnetic fields .
it is also possible to construct the electromagnetic four - potential which depends on gauge transformations .
thus , it is not difficult to guess weinbeg s state vectors are for the electromagnetic field , while matrix a of fig .
[ annals ] is for gauge transformations applicable to the four - potential @xcite . with this point in mind , i published in 1981 a paper with han saying that matrix a performs a gauge transformation @xcite .
we considered a momentum - preserving transformation by considering one rotation followed by two boosts , as shown in fig .
[ loops ] .
we submitted this paper to the american journal of physics instead of the physical review , because we felt that we were not the first ones to observe this .
indeed , in 1972 , kuperzstych got essentially the same result , as indicated also in fig .
[ loops ] .
it is remarkable that he got this result without making reference to wigner s 1939 paper .
he concluded his paper saying that the concept of spin could be generated from his momentum - preserving transformation .
it is remarkable that he derived this result without relying on the concept of the little groups spelled out in wigner s 1939 paper @xcite . in 1953 ,
inn and wigner published a paper on group contractions @xcite .
we can study the contraction of the @xmath0 group to @xmath2 by considering a sphere for @xmath0 and a two - dimensional plane for @xmath2 .
we can then consider this plane tangent to the sphere at the north - pole .
if the radius of the sphere becomes very large , the spherical surface at the north - pole becomes flat enough to accommodate the @xmath2 symmetry .
we can constrauct a flat football field on the surface of the earth .
thus , the @xmath2-like little group for massless particles can be obtained from the @xmath0-like little group for massive particles .
then , what is the physics of the large - radius limit ? in his 1939 paper
@xcite , wigner considered the little group of the massive particle at rest .
what is then the little group for a particle moving along the @xmath1 direction .
the answer is very simple .
it is a lorentz - boosted rotation matrix .
what happens when the momentum becomes infinite ? in order to address this question , we can start from einstein s @xmath116 .
we all know the form of this relation in the limit of small @xmath117 .
we also know the form for the large-@xmath117 limit . with this point in mind , with han and son
, i published a paper telling that the the rotational degrees of freedom around @xmath12 and @xmath13 directions become one gauge degree of freedom while the rotation around the @xmath1 axis remains as the helicity degree of freedom , as @xmath117 becomes infinite @xcite .
after several stages of refinements , we published table [ gen11 ] in the journal of mathematical physics in 1986 @xcite .
wigner liked this table .
this is precisely the reason why i was able to publish a number of papers with him .
however , wigner pointed out to me that the geometry of the two - dimensional plane can not explain the gauge transformation , as indicated in fig .
[ cylin ] .
we thus worked hard to figure out the solution to this problem . for a given sphere
, we can consider also a cylinder tangential to the equatorial belt , as well as a plane tangential to the north pole , as illustrated in fig .
[ cylin ] .
we published this result in the journal of mathematical physics in 1987 @xcite , and another paper in 1990 @xcite .
lawrence biedenharn was the chief editor of the journal .
he was very happy to publish these papers and congratulated me on reactivating wigner s 1939 paper .
i am very happy to include in this report biedenharn s photo with wigner in fig .
[ biedenwein ] , which i took with my canon ae camera in 1988 . included in the same figure is a photo of weinberg talking to wigner while he was a graduate student at princeton in 1957 .
dieter brill contributed this photo .
weinberg was sam treiman s first phd student and got his degree in 1957 .
since i went to princeton in 1958 , i did not have overlapping years with weinberg , but i had to read his thesis to copy his style of writing .
i still like his english style .
of course , i am proud of working with wigner during his late years . on the other and , could i do this job without my own background ?
i had to fix up wigner s work in order to put my own physics program on a solid theoretical ground . when i was graduate student , and for several years after my phd degree , i lived in the world where the origin of physics is believed to be in the complex plane of the s - matrix , and bound states should be described by poles in the complex plane .
in 1965 , when i pointed out those poles do not necessarily lead to localized wave functions @xcite , i had to face a stiff resistance from the influential members of the american physics community .
i choose not to mention their names .
they told me wave functions have nothing to do with physics .
this is a totally destructive attitude .
however , i took their reactions constructively .
we do not know how to lorentz - boost bound - state wave functions , while the s - matrix is a product of the lorentz - covariant field theory .
cccc & & & + + & & & + & massive , slow & covariance & massless , fast + & & & + energy- & & einstein s & + momentum & @xmath3 & @xmath4 & @xmath5 + & & + internal & @xmath6 & & @xmath6 + space - time & & wigner s & + symmetry & @xmath7 & little group & gauge transformations + & & + & & + moving & gell - mann s & covariant & feynman s + h - atom & quark model & bound state & parton model + & & + + thus , my problem was to find at least one wave function that can be lorentz - boosted .
i then realized that dirac in 1945 @xcite and yukawa in 1953 @xcite constructed a gaussian form that can be lorentz - boosted .
in april of 1970 , at the spring meeting of the american physical society , feynman gave a talk where he repeatedly mentioned wave functions .
he was talking about hadrons as bound states of quarks .
my colleagues were saying feynman was absolutely crazy , but he was a savior to me .
let us face the difficulty of boosting wave functions . in 1971
@xcite , with his students , feynman published a paper based on his 1970 talk @xcite .
there , they start with a lorentz - invariant differential equation for particles bound together by harmonic - oscillator potential . however , they produce solutions containing the gaussian form @xmath118 this form is invariant under lorentz transformations , but it is not possible to physical interpretations to @xmath69 variable . on the other hand , feynman s differential equation also produces the solutions containing thee gaussian form @xmath119 in this case , the wave function is normalizable in all the variables .
the form is not invariant .
this means the wave function appears differently to moving observers .
figure [ quapar ] illustrates how differently the wave function look differently . with marilyn
noz , i used the gaussian form of eq.([cov11 ] ) to show that gell - mann s quark model and feynman s parton model are two different limits of one lorentz - covariant entity , and submitted our result to physical review letters .
the referee was very competent and honest .
he / she said he / she really learned what the parton model is all about and the result is important .
however , he would `` enthusiastically '' recommend publication in the comments and addenda section of the phys .
d , instead of prl . we accepted his / her recommendation and published paper as a comment @xcite .
however , what aspect of the fundamental symmetry does this quark - parton reflect ? in order to answer this question , i had to study wigner s 1939 paper , and show that the lorentz - covariant oscillator wave functions are representations of wigner s @xmath0-like little group @xcite .
my continued efforts led to a prl paper of 1989 @xcite . in that paper ,
i expanded table [ gen11 ] to table [ gen22 ] .
this paper also contains the major portion of fig .
[ quapar ] . the elliptic squeeze described in this fiigure
is called the space - time entanglement in the current literature @xcite .
let me summarize what i said above .
many people told me i am totally isolated from the rest of the physics world while working on the problem nobody else worries about .
i disagree .
i have been in touch with all essential physicists in this field , including eugene paul wigner .
in other words , i am qualified to write this report . a. s. wightman _ linvariance dans la mecanique quantique relativiste _ , relations de dispersion et particules elementaires , edited by c. dewitt and r. omnes , ( john wiley & sons inc new york 1962 ) .
pp 150 - 226 .
l. michel _ invariance in quantum mechanics and group extension _ group theoretical concepts and methods in elementary particle physics , lectures of the istanbul summer school of theoretical physics , edited by feza gursey ( gordon and breach , new york , 1962 ) .
pp 135 - 200 . | in 1963 , eugene paul wigner was awarded the nobel prize in physics for his contributions to the theory of the atomic nucleus and the elementary particles , particularly through the discovery and application of fundamental symmetry principles .
there are no disputes about this statement . on the other hand
, there still is a question of why the statement did not mention wigner s 1939 paper on the lorentz group , which was regarded by wigner and many others as his most important contribution in physics . by many physicists ,
this paper was regarded as a mathematical exposition having nothing to do with physics . however , it has been more than one half century since 1963 , and it is of interest to see what progress has been made toward understanding physical implications of this paper and its historical role in physics .
wigner in his 1963 paper defined the subgroups of the lorentz group whose transformations do not change the four - momentum of a given particle , and he called them the little groups .
thus , wigner s little groups are for internal space - time symmetries of particles in the lorentz - covariant world .
indeed , this subgroup can explain the electron spin and spins of other massive particles .
however , for massless particles , there was a gap between his little group and electromagnetic waves derivable maxwell s equations .
this gap was not completely removed until 1990 .
the purpose of this report is to review the stormy historical process in which this gap is cleared .
it is concluded that wigner s little groups indeed can be combined into one lorentz - covariant formula which can dictate the symmetry of the internal space - time time symmetries of massive and massless particles in the lorentz covariant world , just like einstein s energy - momentum relation applicable to both slow and massless particles .
* eugene paul wigner s nobel prize * y. s. kim + department of physics , university of maryland , + college park , maryland 20742 , u.s.a . + e - mail : [email protected] + |
the computation of displacement or stress fields of dislocations in uniform motion @xcite has attracted a lot of attention in the past since the pioneering works of frank @xcite , eshelby @xcite , bullough and bilby @xcite , and weertman @xcite .
the theoretical possibility of supersonic dislocations @xcite was put forward very early @xcite , but had to wait until atomistic simulations to earn some credence @xcite .
although experimental evidence is still lacking , the quest for supersonic dislocations in metals has since gained impetus with the help of current experimental and simulation tools @xcite . however , experimental evidence for supersonic dislocations is already available in systems other than metals , e.g. , in dust plasma crystals @xcite or in relation with seismological events @xcite .
we mainly deal hereafter with two classes of rectilinear dislocations @xcite .
the first one is that of elementary ` point - like ' volterra dislocations .
the other one is that of somigliana dislocations @xcite ( sometimes called _ smeared - out _ dislocations ) .
hereafter we take them with a flat core of finite width , of the specific functional kind that solves both the peierls - nabarro equation @xcite and the weertman equation @xcite with the frenkel ( sine ) pull - back force @xcite .
eshelby @xcite first used this particular dislocation model to study the width of a moving dislocation .
afterwards , it was employed in further studies on motion @xcite . for conciseness , and for lack of a definite name
, such a dislocation will tentatively be called hereafter an _
eshelby dislocation_. a precise definition of it is given in sec .
[ sec : eshdis ] below .
results for volterra dislocations can be obtained as limits of results for somgliana dislocations by letting the core width go to zero @xcite .
such limits define in general distributions or pseudo - functions , and this aspect of volterra dislocations has recently been emphasized as it is particularly important in dynamics @xcite .
more classically , it has long been known @xcite that volterra supersonic expressions involve mach cones in the form of dirac measures along cone generators @xcite . because dirac measures ( or pseudo - functions , for that matter ) can not be plotted in any meaningful way , numerical field calculations in the supersonic regime require considering somigliana ( or other kinds of smeared - out dislocations ; e.g. , @xcite ) instead of volterra dislocations .
somigliana dislocations can be obtained by convolution of the volterra distributional solution with the appropriate dislocation density function and result in more regular fields @xcite .
as pointed out by eshelby @xcite and weertman @xcite in the context of isotropic elasticity , mach cones generated by such smeared - out dislocations are spread over a finite width of order the dislocation width ; thus , they can be evaluated numerically @xcite .
our emphasis is on dislocation motion in anisotropic media .
an anisotropic medium sustains three wave vector - dependent bulk wavespeeds @xcite .
relatively to the glide direction , those wavespeeds define in non - degenerate cases three particular limiting velocities to be compared to the dislocation velocity @xmath0 , hereafter denoted as @xmath1 where the subscripts stand for `` lower '' , `` intermediate '' and `` upper '' , which can be computed from sections of slowness surfaces transverse to the dislocation line , as explained by lothe in refs .
, @xmath2 and @xmath3 are denoted , respectively , @xmath4 ( also @xmath5 in @xcite ) , @xmath6 , and @xmath7 . ] following lothe s terminology ( also , @xcite ) , the inequality @xmath8 defines the _ subsonic _ range ; there are three isolated _ transonic _ velocities @xmath9 , and the _ fully supersonic _
range is @xmath10 .
the qualificative _ supersonic _ applies to all velocities @xmath11 .
we shall call _ intersonic _ the intermediate supersonic ranges @xmath12 and @xmath13 .
is often termed _ transonic _ ( but also _ intersonic _ ) and the range @xmath14 is termed _ supersonic _ , where @xmath15 and @xmath16 are the shear and dilatational wavespeeds , respectively . ] as is well - known , stroh @xcite devised a powerful method to deal with subsonic volterra dislocations of arbitrary character in an anisotropic medium ( see @xcite for reviews ) .
the method requires solving a @xmath17 eigenvalue problem an inexpensive task with today s computers .
moreover , the current formulation of the stroh theory provides basic insight on anisotropic supersonic solutions @xcite .
for instance , the limiting velocities can be determined , as well as the opening angle of the mach cones .
up to three mach cones can arise ( in the fully supersonic regime ) , involving two waves each @xcite .
to our knowledge however , a general formula for the intensity of the mach cones ( the prefactor of the dirac measures , for volterra dislocations ) has not been derived .
besides , no anisotropic supersonic somigliana solution is available .
this prompts us to seek general analytical means to compute supersonic dislocation fields in anisotropic media . on another issue , the geometric configuration of mach cones in anisotropic media can prove surprising .
indeed , payton @xcite reported from a particular solution for the fields of a volterra dislocation in a transversely anisotropic medium , the possible existence of ` backwards ' mach cones , i.e. , v - shaped waves with cone aperture directed _ towards _ the direction of motion .
no explanation for this counter - intuitive phenomenon has been given so far in terms of general principles .
having derived his solution for a uniformly - moving dislocation as the long - time steady - state asymptotics of a transient causal field solution , payton emphasized the necessity of accounting for causality in field expressions to investigate mach cones .
this stems from the fact that mach cones are built over time as caustics of expanding wavefront sets , which is a causal process .
however , the classical exposition of stroh s formalism rests on postulating _ a priori _ functional expressions for the fields , independently of any consideration of causality @xcite .
but without causality , a - causal spurious wavepackets solutions of the field equations could arise , which must be prevented .
accordingly , the purpose of this paper is : first , to endow the stroh formalism with causal properties , which is done in section [ sec : green ] starting from the green function of the wave equation ; second , to derive from this modified formalism field solutions relative to somigliana dislocations ( section [ sec : eshelby ] ) .
the volterra limit of vanishing core size is also examined , revealing the features of mach cones ( geometry , and intensity ) , and a simple existence criterion for payton s ` backward ' cones is derived ( section [ sec : machcones ] ) .
in addition , full - field calculations are compared with the straightforward huygens construction of mach cones from ray surfaces to validate our analytical method of handling fields in the supersonic regime .
a concluding discussion closes the paper ( section [ sec : concl ] ) .
a few technical calculations are developed in [ sec : distr ] and [ sec : appflat ] .
our conventions for the fourier transform of a space- and time - dependent function @xmath18 are as follows : @xmath19 where unless otherwise stated integrations with respect to the space variable @xmath20 and the wave vector @xmath21 are over the @xmath22-dimensional space @xmath23 ( @xmath24 or @xmath25 ) , and integrations with respect to the time @xmath26 and the angular frequency @xmath27 are over @xmath28 .
bold and sans - serif typefaces are used , respectively , to denote vectors ( @xmath29 ) and dyadic tensors ( @xmath30 ) of components @xmath31 . a ` blackboard ' typeface is used to denote fourth - order tensors ( @xmath32 ) of components @xmath33 .
this section briefly reviews causality issues for the green function ( e.g. @xcite , and references therein ) of the anisotropic wave equation , a topic previously addressed by budreck @xcite . a means of accounting for causality in the stroh representation of the green function for uniformly - moving sources is proposed .
the tensor green functions @xmath34 associated with the material displacement in the homogeneous medium are defined as solutions of the inhomogeneous wave equation with unit impulsive body force @xmath35 where @xmath36 is the wave operator @xmath37 defined in terms of the material density @xmath38 , and of the anisotropic elasticity tensor @xmath39 of components @xmath40 . in the fourier representation with wave vector @xmath21 and angular frequency @xmath27 , equation ( [ eq : gdefeqxt ] )
reads @xmath41 different green functions are distinguished by boundary conditions at infinity , and are built on the following template where we introduce the acoustic tensor @xmath42 and the unit @xmath43 matrix @xmath44 : @xmath45^{-1}.\end{aligned}\ ] ] its poles are solutions of the dispersion equation @xmath46 , where @xmath47.\end{aligned}\ ] ] the physical solutions stem from prescribing a way to shift the poles off the real @xmath27-axis @xcite .
the retarded green function of the wave equation ( denoted with a plus superscript ) , and the advanced one ( denoted with a minus superscript ) are defined as @xmath48 where the last equality introduces the notation @xmath49 . the positive quantity @xmath50 represents the reciprocal of some attenuation time of the waves . in lossless media @xmath50 is infinitesimal as in eq .
( [ eq : gdef ] ) , which implements the outgoing ( respectively , incoming ) radiation wave condition in @xmath51 ( resp . , @xmath52 ) .
the advanced green function @xmath52 is useful in the computation of wave intensities and energies .
the following identities hold : @xmath53 average ( @xmath54 ) and radiation ( @xmath55 ) parts of the retarded and advanced green functions are introduced as [ eq : statrad ] @xmath56 so that @xmath57 the average @xmath54 is another green function , since it evidently obeys eq .
( [ eq : gdefeqxt ] ) .
the radiation part @xmath55 is not a green function , but a solution of the homogeneous wave equation @xmath58 ( i.e. , a wavepacket ) .
definition ( [ eq : draddef ] ) , introduced by dirac , is equivalent to computing @xmath55 as the difference between outgoing and incoming fields from / on the source @xcite .
schwinger uses a decomposition such as ( [ eq : gpm ] ) for the retarded field , and interprets @xmath54 ( which changes sign upon inverting the sign of time ) as a reactive part which leads , upon computing powers , to inertial storage of energy in the field by the moving source ; and the part @xmath59 , which keeps its sign under time inversion , as an irreversible radiative part leading to energy dissipation through resistive power @xcite . in relation with green s functions ,
@xmath55 is sometimes referred to as the _ propagator _
( e.g. , @xcite , where it is denoted by @xmath60 ) .
we use this denomination hereafter .
the functions @xmath51 and @xmath55 are simply related through the relationship @xcite [ eq : connectall ] @xmath61 which , combined with eqs . ( [ eq : statrad ] ) , entails the subsidiary relations @xmath62 equation ( [ eq : connect ] ) is an instance of duhamel s principle of building solutions of an inhomogeneous problem from solutions of an homogeneous initial - value ( i.e. , cauchy ) one .
the function @xmath55 is subjected to initial conditions @xmath63 by substitution , one verifies using ( [ eq : init ] ) that @xmath64 in eqs .
( [ eq : connectall ] ) obey the inhomogeneous wave equation ( [ eq : gdefeqxt ] ) .
however , whereas @xmath54 is non - causal ( non - physical ) , @xmath55 is a physically admissible freely - moving wave packet , of a form determined by conditions ( [ eq : init ] ) .
the causal function @xmath51 is retrieved either by combining them according to ( [ eq : gpm ] ) , or by restoring causality in a multiplicative way by means of duhamel s principle ( [ eq : connect ] ) . the green functions @xmath65 and @xmath54 , as well as @xmath55 , are distributions @xcite , whose explicit expressions for an anisotropic homogeneous medium are now reviewed .
we write @xmath66 , where @xmath67 is the unit director of @xmath21 , and @xmath68 is the acoustic operator of components @xmath69 it is symmetric and admits the diagonalization @xmath70 where the three real and positive eigenvalues are squares of ( phase ) wavespeeds @xmath71 for plane elastic waves that propagate in direction @xmath72 .
the projectors @xmath73 are built from associated polarization eigenvectors @xmath74 that form a complete basis .
thus , @xmath75 definitions ( [ eq : fgdef ] ) and ( [ eq : gdef ] ) entail @xmath76 henceforth , and unless otherwise stated , we drop for brevity the dependence on @xmath72 of @xmath77 and @xmath73 .
since @xmath78 , using the sokhotski - plemelj formula ( [ eq : sk ] ) yields eq .
( [ eq : gdef ] ) in distributional form as @xmath79,\end{aligned}\ ] ] where ` @xmath80 ' is a principal - value prescription and @xmath81 is the dirac distribution . substituting ( [ eq : gfourdistr ] ) into definitions ( [ eq : statrad ] ) , one deduces the fourier forms of @xmath82 and @xmath83 as @xmath84 using ( [ eq : cbas].2 ) and eq .
( [ eq : decompdir ] ) below , eq .
( [ eq : dkom ] ) immediately entails @xmath85 which express the initial conditions ( [ eq : init ] ) in the fourier representation . in this paper
, we consider only the two - dimensional ( 2d ) problem , for which we now obtain @xmath55 in space - time coordinates .
fourier inversion of @xmath86 is carried out by integrating over @xmath27 first .
thus , from ( [ eq : dkom ] ) and the expansion @xmath87 = \frac{1}{2c_\alpha k}\left[\delta\left(\omega - c_\alpha k\right)-\delta\left(\omega+c_\alpha k\right)\right],\end{aligned}\ ] ] we deduce @xmath88,\end{aligned}\ ] ] where @xmath89 is the polar angle for the unit director @xmath72 .
its reference orientation for @xmath90 is irrelevant for the time being ( a particular choice will be made in the next section ) .
angular integrals over the unit circle are obviously unchanged by the substitution @xmath91 , which we use in the rightmost exponential , thereby transforming the difference of exponentials within brackets into @xmath92 $ ] . by using next @xmath93 and denoting angular averages by the following shorthand notation @xmath94
, we arrive at the integral representation @xmath95 doubling the integrand and again exploiting the @xmath91 invariance of @xmath77 and @xmath73 turns it into @xmath96 ^ 2}\right\rangle}_{\!\!\phi}}.\end{aligned}\ ] ] combining ( [ eq : dint2d ] ) and ( [ eq : dint2dbis ] ) , and ( [ eq : connect ] ) eventually yields @xmath97 ^ 2}\right\rangle}_{\!\!\phi}}.\end{aligned}\ ] ] mura provides a three - dimensional version of the intermediate expression in ( [ eq : gint2d ] ) , see eq .
( 9.23 ) p. 61 in @xcite .
furthermore , tewary @xcite gives analogous representations of the 3d time - dependent green function based on duhamel s principle , and radon transforms @xcite .
however , for 2d problems radon - transform and fourier methods such as above are identical @xcite .
using the mutual orthogonality of the eigenvectors , one verifies that @xmath98 in the form ( [ eq : dint2d ] ) solves the homogeneous wave equation @xmath99 , in which gradients must now be considered as two - dimensional with respect to @xmath100 .
moreover , the initial - value conditions ( [ eq : init ] ) must hold with @xmath101 replaced by the two dimensional @xmath102 . to check
, we note first that , obviously , @xmath103 by symmetry .
next , differentiating ( [ eq : dint2d ] ) with respect to time we get ( with a finite - part prescription , hereafter denoted by ` @xmath104 ' ) @xcite . ]
@xmath105 ^ 2}\right\rangle}_{\!\!\phi}}.\end{aligned}\ ] ] taking @xmath106 in this expression allows one to retrieve the expected condition @xmath107 , on account of the completeness relation ( [ eq : cbas])@xmath108 , and of the interesting distributional identity ( [ sec : distr ] ) @xmath109 the integral form ( [ eq : dint2dbis ] ) of @xmath98 is now computed exactly by means of stroh s formalism following hirth and lothe s notations and conventions @xcite .
the formal connection between the dynamic kernels and those showing up in the problem of sources moving at constant velocity is revealed by introducing a fictitious ` velocity ' variable @xmath110 @xcite , and modified elastic constants @xcite @xmath111 for two real vectors @xmath29 and @xmath112 we abbreviate by @xmath113 @xcite the @xmath114 matrix of components @xmath115 due to the minor voigt symmetry of the elasticity tensor @xmath116 , different indexing conventions for the @xmath117 term in eq .
( [ eq : modc ] ) are found in the literature ( e.g. , @xcite ) .
the bracket notation ( [ eq : hlbrack ] ) must be defined consistently with the chosen convention so as to reproduce the results below . from ( [ eq : modc ] ) , ( [ eq : hlbrack ] ) and the eigenvalue decomposition ( [ eq : diag ] ) , ( [ eq : cbas ] ) one deduces that @xmath118=\rho\sum_{\alpha=1}^3 \left[c^2_\alpha({\mathbf{\widehat{k}}})-({\mathbf{\widehat{k}}}\cdot{\mathbf{v}})^2\right]{\mathsf{p}}^\alpha({\mathbf{\widehat{k}}}),\end{aligned}\ ] ] which yields the inverse tensor @xcite ( eq . ( 194 ) in that reference ) @xmath119 substituting this expression into eq .
( [ eq : dint2dbis ] ) then yields the propagator in the form @xmath120 rotating bases .
( a ) as used in sec .
[ sec : constroh ] .
( b ) as used in sec .
[ sec : eshdis].,width=264 ] this expression can be evaluated by means of the stroh formalism .
indeed , introduce the unit vector @xmath121 and identify @xmath72 with the complementary orthogonal vector @xmath122 ( fig .
[ fig : fig1]a ) ; note that we do not assume here that @xmath123 .
then , one recognizes in ( [ eq : davexpr ] ) one of the angular averages that enter the well - known barnett - lothe integral version of stroh s theory @xcite .
the latter is briefly summarized hereafter ( see , e.g. , @xcite for details ) .
it hinges on using the @xmath124 nonsymmetric matrix @xcite @xmath125 the associated eigenvalue problem @xmath126 involves the right eigenvector @xmath127 , which defines from its components ordered as @xmath128 @xmath129 two 3-vectors @xmath130 and @xmath131 that correspond to polarizations of the displacement , and traction vectors in the plane @xmath132 , respectively .
the matrix @xmath133 has six eigenvalues @xmath134 , @xmath135 and associated eigenvectors @xmath136 .
these vectors are normalized such that @xmath137 for @xmath138 , the eigenvalues constitute three pairs of conjugate complex numbers conventionally labelled so that @xmath139 for @xmath140 , and such that @xmath141 ( the overbar denotes the complex conjugate ) . from ( [ eq : eigsyst ] ) , the same property is inherited by @xmath142 and @xmath143 , and @xmath144 for @xmath140 .
a key property of the formalism is that both vectors @xmath145 and @xmath146 are independent of the orientation angle @xmath89 .
however , @xmath134 depends on it as @xmath147 where @xmath148 is a complex constant .
this expression shows that @xmath149 and @xmath150 are of same signs @xcite . as @xmath0 increases , two
@xmath134s become real - valued , as well as their associated angles @xmath148 , each time @xmath0 crosses one of the transonic velocities . in the supersonic range
all the @xmath134 are real .
this point is further discussed in sec .
[ sec : machcones ] in connection with mach cones .
an important quantity to be used hereafter is the angular average @xmath151 , defined as a principal value .
whenever @xmath148 has a nonzero imaginary part , the integral over angles is nonsingular , with the result @xmath152 , where the sign is that of @xmath150 @xcite ( by contour integration on the unit circle with integration variable @xmath153 ) .
by contrast , when @xmath148 is real the integral is divergent at @xmath154 mod .
@xmath155 , and the indispensable ` @xmath80 ' prescription makes it finite by handling these singularities as principal values . using ( [ eq : ptan ] ) , one easily gets from an obvious change of variables the result @xmath156 .
thus , introducing @xmath157 we have @xmath158 expanding the matrix equation ( [ eq : eigsyst ] ) shows that the vectors @xmath143 are connected to the @xmath142 by the relationship @xmath159\cdot{\mathbf{a}}^\alpha,\end{aligned}\ ] ] and that the existence of nonzero eigenvectors @xmath142 requires the eigenvalues @xmath160 to be determined in terms of @xmath0 by the equation @xmath161+p^2(nn)\right\}&=0.\end{aligned}\ ] ] as a consequence , @xmath133 can be shown to admit the decomposition @xmath162 identifying expressions ( [ eq : ndef ] ) and ( [ eq : ndecomp ] ) , decompositions in the form of sum rules of the block matrices that make up @xmath133 follow ; notably , @xmath163 moreover , the following closure relations hold @xmath164 we can now compute @xmath98 by substituting expression ( [ eq : nnm1 ] ) into eq .
( [ eq : davexpr ] ) , which results in @xmath165 emphasizing the functional dependence of the vectors , and using ( [ eq : palphaav ] ) one concludes that @xmath166 where the sum is restricted to indices of eigenvalues with strictly positive imaginary part , and where complex - conjugation properties have been used .
apart from different sign and normalization conventions , equ .
( [ eq : dstroh ] ) is equivalent to wu s equations ( 3.1213 ) @xcite .
although expression ( [ eq : dstroh ] ) is sometimes referred to as a green s function ( e.g. , @xcite ) , the above derivation clarifies its nature as a propagator .
the green functions @xmath167 stem from multiplying ( [ eq : dstroh ] ) by the appropriate time - dependent prefactor , as in eqs .
( [ eq : connectall ] ) . up to now , the vector @xmath168 was a shorthand for @xmath169 . in the rest of the paper
, it will denote a ` true ' velocity .
so , the dislocation moves at _ constant _ velocity @xmath170 , where the scalar @xmath0 can be of any sign .
the unit vector @xmath171 , such that @xmath172 is the slip - plane normal .
the dislocation line is oriented along @xmath173 , and the plane spanned by @xmath174 and @xmath171 is the co - called _
sagittal plane_. then , the components of the source tensor are all of the type @xmath175 , of fourier transform @xmath176 .
the induced field components all have the following form , where the integral on @xmath177 is over all @xmath22-dimensional space : @xmath178 where the last line is a well - known representation of the field in terms of inverse fourier transforms . in this expression ,
the fundamental quantity is @xmath179 , namely , @xmath180 hereafter , motion in anisotropic media will be addressed on the basis of formula ( [ eq : invfour ] ) . to this aim
, we need to write @xmath179 in a form computable from the stroh formalism .
this is possible only if we can account for the prescription @xmath181 by modifying the elastic tensor _ in a way that does not depend on @xmath72 _ ; otherwise , the crucial property that @xmath142 and @xmath143 do not depend on @xmath89 ( see previous section ) would be destroyed .
a convenient way to do this the central idea of the paper consists in shifting the algebraic velocity by a small positive imaginary quantity , thus introducing the complex velocity vector @xmath182 where @xmath183 is an infinitesimal positive number of irrelevant exact magnitude .
then , assuming @xmath184 , @xmath185 accordingly , with @xmath168 now the true velocity , we modify senz s velocity - dependent ` elastic constants ' ( [ eq : modc ] ) into the complex - valued elastic moduli @xmath186 further introducing @xmath187 , it follows that @xmath188\delta_{il}.\end{aligned}\ ] ] upon diagonalizing this expression in the polarization basis , we obtain @xmath189.\end{aligned}\ ] ] furthermore introducing the function @xmath190 il follows that @xmath191 since @xmath192 , comparing ( [ eq : limfkv ] ) with ( [ eq : gkkv ] ) shows that @xmath193{\mathsf{f}}^+_{\epsilon}({\mathbf{k}},{\mathbf{v}}).\end{aligned}\ ] ] one deduces from definitions ( [ eq : gokom ] ) and ( [ eq : dkom ] ) and the above that [ eq : g0dkv ] @xmath194 as will be made clear in the following , the stroh formalism can now be used to compute @xmath195 , thanks to identity ( [ eq : nnm1 ] ) .
the green functions @xmath196 are then retrieved by combining expressions ( [ eq : g0dkv ] ) according to ( [ eq : gpm ] ) .
to summarize , the causal green function relevant to a source in uniform motion can not be directly obtained from the stroh formalism because of the factor @xmath197 in ( [ eq : gwithsign ] ) . however ,
taken separately , its real ( reactive ) and imaginary ( radiative ) parts can be , to be reassembled afterwards to retrieve @xmath198 .
our assumption that @xmath199 does not impair the generality of the above derivation , since the dirac contributions to the green function ( [ eq : gkkv ] ) vanish anyway if @xmath200 .
this section is devoted to obtaining the distortion and stress fields produced by a uniformly moving somigliana dislocation , and further specialized into a dislocation of the eshelby type .
the somigliana dislocation is represented by the plastic eigenstrain tensor @xmath201 , which constitutes the source of the elastodynamic fields in the green s function approach @xcite .
the dislocation , moving at constant velocity @xmath168 , is such that in the direct and fourier representations ( respectively ) , [ eq : betaexpr ] @xmath202 its shape , assumed rigid because of uniform motion , is completely characterized by @xmath203 , or by @xmath204 in the fourier representation .
the material displacement induced by the dislocation in the surrounding medium is of the form ( [ eq : invfour ] ) , and reads @xmath205 introducing generic response kernels @xmath206 of components in ( [ eq : bline1 ] ) and reorganizing terms ( @xcite , p. 351 ) : @xmath207g_{jp}({\mathbf{k}},\omega ) = ( \epsilon_{oin}\epsilon_{omk } k_n k_q c_{pqml}-\rho\,\omega^2\delta_{ik}\delta_{pl})g_{jp}({\mathbf{k}},\omega).\end{aligned}\ ] ] ] @xmath208 the elastic distortion @xmath209 reads @xmath210 decomposition ( [ eq : gpm ] ) of the green function into reactive and radiative parts induces a similar additive decomposition of @xmath211 that carries over to the fields .
hereafter , the reactive and radiative parts of @xmath212 are denoted with a superscript @xmath213 or @xmath83 , respectively .
explicitly , we shall write @xmath214 where [ eq : bline ] @xmath215 from which @xmath216 and @xmath217 are determined by expressions akin to ( [ eq : betab ] ) . the stress follows from the generalized hooke law as @xmath218 .
we now specialize to the eshelby dislocation . as recalled in the introduction , it is a natural solution of the ( static ) peierls - nabarro and ( steady - motion ) weertman models for the dislocation core shape . as it has been used as well to represent
non - uniformly moving dislocations @xcite , it is of special theoretical interest .
the eshelby dislocation has a core density function per unit burgers vector of a simple ( lorentzian - type ) analytic structure , namely @xmath219 where , to avoid dragging along factors @xmath220 , the quantity @xmath221 is introduced as the _ half core width _ rather than as the core width .
this density depends solely on the in - plane cartesian coordinate @xmath222 in the direction of motion .
we introduce the co - moving position vector @xmath177 and abscissa @xmath223 defined as @xmath224 the dislocation is flat with respect to the out - of - plane coordinate @xmath225 .
the functions @xmath226 define a @xmath81-sequence such that @xmath227 . in this limit @xmath228 , the eshelby dislocation reduces to a volterra dislocation .
the associated plastic eigenstrain is localized on the slip plane , and reads @xmath229 equation ( [ eq : disc ] ) is our definition of the eshelby dislocation .
the slip discontinuity @xmath230 spans the whole real axis .
it reduces to @xmath231 in the volterra limit @xmath228 .
conversely , eq . ( [ eq : disc ] ) is retrieved by taking the convolution of the latter volterra expression by the density ( [ eq : coreflat ] ) . for further purposes , we note that @xmath232 since in the fourier domain @xmath233 we deduce from the fourier transform of @xmath234 that @xmath235 we begin by computing the reactive distortion @xmath216 from ( [ eq : bline0 ] ) .
it reads @xmath236\frac{n_k { \mathrm{e}}^{{\mathrm{i}}{\mathbf{k}}\cdot{\mathbf{r}}'-a k|{\mathbf{\widehat{k}}}\cdot{\mathbf{m}}|}}{{\mathbf{k}}\cdot{\mathbf{m}}+{\mathrm{i}}0^+}.\end{aligned}\ ] ] the term within square brackets has zero degree of homogeneity in @xmath21 , and thus does not depend on the modulus @xmath237 .
the integral over @xmath237 can therefore be done right away , resulting in the distribution @xmath238 to deal with the remaining angular integral over @xmath72 in ( [ eq : beta0 ] ) , one introduces a rotated orthonormal basis @xmath239 , with components @xmath240 and @xmath241 in the fixed basis @xmath242 ( fig . [
fig : fig1]b ) . by periodicity , the angular integral over the direction @xmath72 in ( [ eq : beta0 ] ) is equivalent to one over @xmath89 .
this brings ( [ eq : beta0 ] ) down to the form @xmath243 from definition ( [ eq : modcmplx ] ) , contractions involving the usual elastic tensor @xmath116 can be expressed in terms of contractions with the modified , complex - valued one , @xmath244 . because @xmath245 , we have @xmath246 which is real - valued .
using this result and ( [ eq : g0kv ] ) brings the numerator in ( [ eq : beta0aaa ] ) in the form @xmath247_{jl}-n_i\delta_{jl}\right).\end{aligned}\ ] ] to carry out the average over @xmath89 , the constant vector @xmath171 is decomposed over the rotated basis @xmath239 as @xmath248 after some simplifications involving the identities @xmath245 and @xmath249 , which hold by definition of the rotated basis , the following expansions obtain : @xmath250 + ( { \mathbf{n}}\cdot{\mathbf{l}})^2(l\,l)_\epsilon.\end{aligned}\ ] ] substituting expressions ( [ eq : nlk ] ) and ( [ eq : expankn ] ) into ( [ eq : bracket0 ] ) yields , after some cancellation of terms , @xmath251_{jl } -l_i\delta_{jl}\right\}\nonumber\\ & = -({\mathbf{\widehat{k}}}\cdot{\mathbf{m}}){\mathop{\mathrm{re}}}\left\{{\widehat{k}}_i[({\widehat{k}}{\widehat{k}})_\epsilon^{-1}\cdot({\widehat{k}}\,l)_\epsilon]_{jl}-l_i\delta_{jl}\right\},\end{aligned}\ ] ] where the identity @xmath252 ( @xmath253 ) has been used . inserting this expression into ( [ eq : beta0aaa ] ) results in @xmath254_{jl}-l_i\delta_{jl}\right\ } }
{ { \mathbf{\widehat{k}}}\cdot{\mathbf{r}}'+{\mathrm{i}}a |{\mathbf{\widehat{k}}}\cdot{\mathbf{m}}|}\right\rangle}_{\!\!\phi}},\end{aligned}\ ] ] where a factor @xmath255 has been eliminated between the denominator ( [ eq : intk ] ) and the numerator ( [ bracket0 ] ) , the prescription @xmath181 being irrelevant in this case .
the real and imaginary parts of the denominator are separated as @xmath256 thanks to this decomposition , we keep in the integrand only the even terms , which do not change their sign under the inversion symmetry @xmath257 ( i.e. , @xmath91 and @xmath258 ) ; the odd ones do not contribute .
thus , expression ( [ eq : beta0angav ] ) takes the form @xmath259_{jl}({\mathbf{\widehat{k}}}\cdot{\mathbf{r}}'){\widehat{k}}_i}{({\mathbf{\widehat{k}}}\cdot{\mathbf{r}}')^2+a^2 ( { \mathbf{\widehat{k}}}\cdot{\mathbf{m}})^2}\right\rangle}_{\!\!\phi } } -\frac{b_l}{2\pi}{{\left\langle \frac{({\mathbf{\widehat{k}}}\cdot{\mathbf{r}}')l_i}{({\mathbf{\widehat{k}}}\cdot{\mathbf{r}}')^2+a^2 ( { \mathbf{\widehat{k}}}\cdot{\mathbf{m}})^2}\right\rangle}_{\!\!\phi}}\delta_{jl}.\end{aligned}\ ] ] invoking next identity ( [ eq : nnnmm1 ] ) to simplify the leading angular integral yields @xmath260 where subscripts have been introduced to emphasize that the quantities @xmath142 , @xmath143 , and @xmath261 now depend on @xmath183 .
expression ( [ eq : beta0flat ] ) reveals that the finite width @xmath221 has a twofold regularizing action on the angular averages : on the one hand , it sets the result to zero at the origin @xmath262 ; on the other hand , it provides a regularization to handle the singularities at @xmath263 .
the volterra limit @xmath264 is examined in sec .
[ sec : machcones ] below . to complete the calculation the following shorthand notations
are used : @xmath265 the main difference with the usual formalism ( sec .
[ sec : constroh ] ) is that because the infinitesimal @xmath183 is always non - zero , we have @xmath266 whatever the dislocation velocity , even though @xmath267 can become real - valued in the limit @xmath268 for supersonic velocities .
this fact proves crucial in determining the mach cone shapes in supersonic regimes , as will be seen in sec .
[ sec : machcones ] below .
the angular integrals in ( [ eq : beta0flat ] ) are carried out componentwise in the @xmath242 basis , taking advantage of the explicit form ( [ eq : ptan ] ) of the functions @xmath269 . to this aim
we introduce the angle @xmath270 such that @xmath271,\qquad { \mathbf{\widehat{k}}}\cdot{\mathbf{\widehat{r}}}'=\sin(\gamma-\phi).\end{aligned}\ ] ] in terms of the cartesian coordinates @xmath272 we obtain thus [ eq : intflat ] @xmath273 inserting these results into ( [ eq : beta0flat ] ) and invoking the completeness relation ( [ eq : caa]@xmath108 ) , we conclude that @xmath274 for definiteness , the calculation of the first component ( i.e. , along @xmath174 ) of integral ( [ eq : intflat1 ] ) by contour integration on the unit circle , as is usual within the barnett - lothe approach is explained in detail in [ sec : appflat ] .
the other integrals in this section are computed likewise .
we turn next to the `` propagator '' part , which by ( [ eq : blined ] ) and ( [ eq : betapflat ] ) , and after use of ( [ eq : intk ] ) , reads @xmath275 from expression ( [ eq : dkv ] ) of @xmath276 , we have @xmath277_{jp}(k_q c_{pqkl}n_k){\mathop{\text{sign}}}({\mathbf{\widehat{k}}}\cdot{\mathbf{m}}).\end{aligned}\ ] ] so that , by substitution into ( [ eq : betad ] ) , @xmath278_{jp}(k_q c_{pqkl}n_k){\mathop{\text{sign}}}({\mathbf{\widehat{k}}}\cdot{\mathbf{m}})}{({\mathbf{\widehat{k}}}\cdot{\mathbf{m}}+{\mathrm{i}}0^+)({\mathbf{\widehat{k}}}\cdot{\mathbf{r}}'+{\mathrm{i}}a |{\mathbf{\widehat{k}}}\cdot{\mathbf{m}}|)}\right\rangle}_{\!\!\phi}}.\end{aligned}\ ] ] employing the same means as above , this expression reduces to @xmath279 the angular integral evaluates to @xmath280.\end{aligned}\ ] ] upon substituting into ( [ eq : betadflat ] ) , we obtain thus @xmath281a_{\epsilon\,j}^\alpha l_{\epsilon\,l}^\alpha b_l.\end{aligned}\ ] ] however , using once more the completeness relation ( [ eq : caa].2 ) , we observe that the first term within brackets is real - valued and does not contribute to the result due to the leading @xmath282 operator .
therefore , @xmath283 eventually , adding eqs .
( [ eq : beta0flat1 ] ) and ( [ eq : betadflat1 ] ) and carrying out a few obvious reorganizations yields the total elastic distortion in the form @xmath284
the distortion components ( [ eq : betaflat ] ) in the volterra limit @xmath228 are easily deduced . to this purpose , it is necessary to distinguish between real and complex roots @xmath285 in the limit @xmath268 . when @xmath285 is such that its limit @xmath286 is real , we get from the sokhotski - plemelj formula ( [ eq : sk ] ) @xmath287 where , because it is defined as a limit , @xmath288 must be considered as nonzero even though the limit @xmath289 is real [ see remark after eq .
( [ eq : shnot ] ) ] . substituting ( [ eq : skp ] ) into ( [ eq : betaflat ] ) , and observing that @xmath290 yields the distributional expression ( where the superscript v on @xmath212 stands for ` volterra ' ) @xmath291 b_l \delta(x'+p^{0\alpha } y).\end{aligned}\ ] ] the principal - value prescription within the first sum is necessary only for the subset of @xmath292 values such that @xmath293 , and can be omitted whenever @xmath289 has a finite imaginary part since the latter prevents the denominator from vanishing .
the ` prime ' restricts the second sum to the same subset of @xmath292 values .
mach cones are characterized as follows .
the rightmost sum in ( [ eq : betavolt ] ) features dirac measures supported by straight lines of equations @xmath294 , which represent the mach cones .
each v - shaped cone is made of two half - lines ( ` branches ' ) , and thus is associated with two different eigenvalues @xmath134 .
selection of one particular cone branch @xmath292 is done via its factor @xmath295 .
for instance , the branch of a mach cone in the upper half - plane @xmath296 is such that @xmath297 , because @xmath298 if @xmath296 and zero otherwise ; the same holds with all signs reversed for the lower branch in the half - plane @xmath299 .
moreover , the polarization scalar product @xmath300 makes the intensity and sign of the elastic fields of each branch depend on the dislocation character .
in particular , the `` ghost branches '' for which this scalar product vanishes are extinguished and do not contribute to the field even though they exist so to say geometrically speaking . in the subsonic range @xmath138 where all the @xmath289 have non - zero imaginary part , eq .
( [ eq : betavolt ] ) reduces to the ` classical ' result @xcite for a volterra dislocation .
indeed , we have then @xmath301 remembering that @xmath302 , our numbering conventions and the complex - conjugacy properties of the eigenvalues and eigenvectors ( see section [ sec : constroh ] ) imply that @xmath303 b_l\nonumber\\ & = \frac{1}{\pi}{\mathop{\mathrm{im}}}\sum_{\alpha=1}^3\frac{(m_i+p^{0\alpha } n_i)}{x'+p^{0\alpha } y}a_{\epsilon\,j}^\alpha l_{\epsilon\,l}^\alpha b_l = \frac{1}{2{\mathrm{i}}\pi}\sum_{\alpha=1}^6 s_\alpha\frac{(m_i+p^{0\alpha } n_i)}{x'+p^{0\alpha } y}a_{\epsilon\,j}^\alpha l_{\epsilon\,l}^\alpha b_l,\end{aligned}\ ] ] which is the classical result .
maln @xcite provided an explanation of mach cones from a geometric construction based on the section of the _ slowness surface _ in the sagittal plane .
as we shall shortly see , the argument is incomplete as it only involves phase velocities and does not allow one to explain payton s ` backward ' cones @xcite(see introduction ) .
the geometrical construction that provides the explanation involves group velocities .
first , we briefly review the definition of slowness and group - velocity surfaces ( see , e.g. , @xcite for details ) , also known as _ ray surfaces _ in the literature ( e.g. , @xcite ) . with definition ( [ eq : omegadef ] ) of @xmath304 , the dispersion equation @xmath305 is solved for the ( positive ) phase velocities @xmath306 as a function of the wave direction @xmath307 in the sagittal plane .
the slowness vectors are @xmath308 .
parametric plots of the endpoints of the vectors @xmath309 , using @xmath72 as a parameter , define slowness surfaces , in which the radius vector lies along @xmath72 .
by contrast , group velocities are defined as @xmath310 parametric plots of the endpoints of the vectors @xmath311 using @xmath72 as a parameter define group - velocity surfaces .
the latter represent the positions at time @xmath312 s of wavefronts emitted from the origin at @xmath106 .
group - velocity surfaces are sometimes also called energy - velocity surfaces. the velocity of energy transport in a plane wave is indeed equal to its group velocity .
group and phase velocities are different from one another in anisotropic media due to directional interference effects , which makes anisotropic media dispersive @xcite .
the vector @xmath313 lies in the direction normal to the slowness surface at @xmath72 and , save for special directions of symmetry , is not directed along @xmath72 .
quite generally , @xmath314 figures [ fig : fig2](a@xmath315 ) and ( b@xmath315 ) represent sections of slowness surfaces in fe , , @xmath316 , @xmath317 , mass density @xmath318 g/@xmath319 , and lattice parameter @xmath320 nm at 298 k @xcite . ] in the sagittal plane of the two different slip systems ( see legend ) . in spite of intersection points ,
the three individual branches of solution are unambiguously identified using the fact that they are smooth functions of @xmath72 .
surfaces ( qs@xmath321 ) in the figure are of quasi - shear character ( shear polarization with admixture of longitudinal polarization ) , and the inner one ( ql ) is quasi - longitudinal ( longitudinal polarization with admixture of shear polarization ) . intersections of slowness curves with vertical lines of abscissa @xmath322 have coordinates @xmath323 , which provides the connection with the stroh formalism @xcite .
such intersections exist only if their associated @xmath289 is real .
this is possible only for @xmath0 above certain bulk limiting velocities of the problem ( open dots in the figure ) for which the vertical lines are tangent to the slowness curves ( represented as dashed in the figures ) .
thus , limiting velocities are @xmath324 , and @xmath325 ms@xmath326 for the slip system in fig .
[ fig : fig2](a@xmath315 ) , and @xmath327 , @xmath328 , and @xmath329 ms@xmath326 for the slip system in fig .
[ fig : fig2](b@xmath315 ) . whereas the generic case , illustrated in fig .
[ fig : fig2](b@xmath315 ) is that of three limiting velocities , the case of fig .
[ fig : fig2](a@xmath315 ) is degenerate since it involves only two velocities .
a similar degeneracy arises in isotropic media where the three slowness curves are perfect circles , with the two shear curves superimposed .
slowness and group - velocity surfaces for fe .
( a@xmath330 ) and ( b@xmath330 ) : sections of slowness surfaces in the sagittal plane @xmath331 ( in units of @xmath332 sm@xmath326 ) .
open dots : limiting values for @xmath322 , where @xmath0 is the source velocity ; ( a@xmath333 ) and ( b@xmath333 ) : sections of group - velocity surfaces in the sagittal plane ( in units of @xmath334 ms@xmath326 ) .
( a@xmath321 ) : slip system @xmath335 $ ] , @xmath336 ( dislocation direction @xmath337 $ ] ) ; ( b@xmath321 ) : slip system @xmath335 $ ] , @xmath338 ( dislocation direction @xmath339$]).,width=453 ] figures [ fig : fig2](a@xmath108 ) and ( b@xmath108 ) represent sections of group - velocity surfaces in fe in the sagittal plane of two slip systems considered .
cusps in group - velocity surfaces arise from non - convex parts in slowness surfaces ( here , in @xmath340 branches ) .
when interpreted in units of meters , group - velocity surfaces provide the shape of wavefronts at @xmath312 s emitted at @xmath106 by a source at the origin of coordinates @xcite .
now , let @xmath341 be this set of group - velocity surfaces centered on the origin .
the well - known huygens construction @xcite for dislocations in isotropic media ( where the situation is simpler because phase and group velocities coincide ) can be generalized to anisotropic media , using group - velocity surfaces to build mach cones as envelopes of the collection of wavefronts emitted over time @xcite .
thus , let @xmath342 be the translation operator that translates a set of vectors by a vector @xmath343 . the collection of wave fronts emitted by a moving dislocation located at @xmath344 at time @xmath345
can then be written as @xmath346 which is the union over time of sets of ray surfaces .
each set is built from @xmath341 , translated by the retarded position vector @xmath347 of the dislocation ( to account for the motion of the emission point of the pulses ) , and expanded homothetically by a factor @xmath348 ( to account for pulse expansion by wave propagation ) .
mach cones are envelopes of this set ( i.e. , caustics , where radiated energy is concentrated ) @xcite . for the uniform motion at velocity @xmath168 considered here one has @xmath349 .
however , the construction , that rests on the concept of retarded fields , holds for non - uniform motion as well @xcite .
as it is only of geometric nature , it provides no information about the intensity of cone branches , some of which may vanish for reasons of polarization as noted in sec .
[ sec : diraclines ] .
the construction is illustrated in fig .
[ fig : fig3 ] where stress components computed from the field formula ( [ eq : betaflat ] ) are displayed for two velocities in a full - field representation ( right ) , together with the corresponding solutions on the slowness surfaces ( left ) .
the velocity in ( a ) , @xmath350 ms@xmath326 is such that one forward and one backward cones are present . in ( b ) , the velocity is higher , and two forward cones are generated . on the full - field plots have been superimposed the huygens construction from the group - velocity surfaces of fig .
[ fig : fig2 ] , as well as cone lines of equations @xmath351 deduced form the theoretical expression ( [ eq : betavolt ] ) .
the former perfectly reproduces the latter in agreement with the full - field plots .
this triple comparison makes clear that the envelope of the cusps of the group - velocity surfaces of fig .
[ fig : fig2 ] does _ not _ determine the mach cone opening angle , as those endpoints do not pertain to any caustic in general : from a mathematical standpoint the cone opening angle is uniquely defined from the above linear equation , which clarifies some ambiguities in recent interpretations of the huygens construction in anisotropic media @xcite .
edge dislocation in fe , with @xmath352 , @xmath353 for the two velocities indicated ( units of @xmath334 ms@xmath326 ) . left : solutions on the slowness surfaces ; right : selected full - field stress components computed from eq .
( [ eq : betaflat ] ) , with the huygens construction superimposed .
the half core width is @xmath354 in ( a ) and @xmath355 in ( b ) @xcite .
distances are measured in units of the interplane distance @xmath22 for this glide system . for better display fields have been cut - off as indicated in the bar legend .
white dashed lines drawn upon the cones are the theoretical loci of cone - branch lines ( see sec .
[ sec : diraclines ] ) . for better display ,
the construction only involves a restricted number of wavefronts built from the quasi - shear branches @xmath356 of fig .
[ fig : fig2](a@xmath108 ) ( the only relevant branches for mach cones , at the velocities considered ) .
( for a color version of the figure , the reader is referred to the web version of this article.),width=453 ] fig .
[ fig : fig3](a ) correlates the existence of backward cones with the fact that one pair of solutions deduced from the slowness surfaces involve normals ( group - velocity vectors ) pointing downwards in the upper half plane , and upwards in the lower half - plane . fig .
[ fig : fig4 ] , which extends maln s construction to account for group - velocity effects @xcite illustrates this in detail .
the angle indicated is the same in the slowness - surface construction ( left ) and the vector construction in the physical space ( right ) .
the half - cone branches are indicated as dashed lines .
the bottom construction shows that while the wave vector @xmath72 points upwards , the group velocity points downwards , and is responsible for the physical backward mach cone branch in the lower half - plane . in both cases ,
the following relation is obeyed , with @xmath357 : @xmath358 geometric constructions for forward and backward mach cones including group - velocity considerations .
( a ) ` normal ' ( forward - cone ) case ; ( b ) ` anomalous ' ( backward - cone ) case . left : constructions from a piece of the section of the slowness surface .
right : corresponding constructions in physical space , with backward ( top ) and forward ( bottom ) .
the resulting half - branches of the mach cones are represented as dashed lines.,width=377 ] examination of eq .
( [ eq : skp ] ) reveals that for this effect to be reproduced by the mach - cone ( dirac ) part in this equation , it is necessary that @xmath359 @xmath360 be equal to the sign of @xmath361 .
this property we demonstrate as follows .
first , by setting @xmath362 , and @xmath363 , .
] it is easily verified that @xmath364 which makes explicit the connection between the dispersion equation , see eq .
( [ eq : omegadef ] ) , and the stroh eigenvalue equation ( [ eq : pv ] ) . note in passing that solutions are such that @xmath365 , in agreement with the geometric constructions of fig .
[ fig : fig4 ] ( left ) where @xmath366 .
now , adding a small imaginary part @xmath367 to @xmath0 ( @xmath368 ) , any real solution @xmath369 of the equation @xmath370 varies by an amount @xmath371 , and @xmath27 and @xmath21 vary by respective amounts @xmath372 and @xmath373 .
thus , @xmath374 then , because @xmath375 and @xmath0 are such that @xmath376 , one deduces that @xmath377 where definition ( [ eq : gvdef ] ) of the group velocity has been used .
so , the infinitesimal imaginary part of @xmath375 generated by the adjunction of @xmath378 to the velocity in the stroh formalism , to account for causality , has indeed the necessary sign . from sections of slowness surfaces , it easy to assess the range of velocities for which backward cones can be present : it is simply the one where there exists normals to the slowness plots , that point towards the horizontal axis .
for instance , in the two cases of fig .
[ fig : fig2 ] , those ranges are comprised between the lower bulk limiting velocity , and the velocity where the two quasi - shear slowness branches qs@xmath321 intersect mutually on the horizontal axis ; namely , @xmath379 in case ( a@xmath315 ) , and @xmath380 in case ( b@xmath315 ) ( in units of @xmath334 ms@xmath326 ) . leaving it to the reader , the same analysis as in sec .
[ sec : diraclines ] could be carried out separately on the reactive and radiative parts of the distortion field , eqs .
( [ eq : beta0flat1 ] ) and ( [ eq : betadflat1 ] ) , respectively .
one would see that , individually , each of those parts produces for faster - than - wave motion x - shaped wavefronts involving a - causal cone branches , instead of v - shaped ones .
however , the unphysical branches vanish by sign compensation in the sum of both terms , eq . ( [ eq : betavolt ] ) .
to conclude , we summarize our results and put them into perspective .
first , we showed that the stroh formalism can be very easily extended to supersonic velocities by means of an analytic continuation to complex values of the algebraic velocity @xmath0 , with infinitesimal positive imaginary component .
this device was discovered in former investigations by pellegrini of the isotropic theory of the equation of motion of dislocations @xcite .
however that first proof stemmed from a somewhat _ ad hoc _ argument , as it resulted from comparing formulas @xcite previously obtained by considering separately the subsonic and supersonic cases . instead , by tracing back the analytic continuation to the radiation condition that defines the anisotropic green functions , the present work gives the method a firm physical basis , and puts it into a far broader perspective , paving the way , e.g. , for studying supersonic regimes in the anisotropic weertman equation .
concretely , it is no more necessary to derive field - related velocity - dependent theoretical expressions separately in the subsonic and supersonic regimes to cover the whole velocity range , as was previously done .
it now merely suffices to compute subsonic expressions , and to continue them by means of the replacement @xmath381 , which is most easily done in numerical computations by making @xmath183 a very small number . in this way
, an expression valid for _ all _ velocities is obtained .
adams @xcite came close to the point in 2001 , in the context of a weertman - type equation for slip - pulse propagation at the interface between two different isotropic media .
he noticed that unique expressions of the coefficients of the governing equation apply indifferently to subsonic or intersonic velocities , provided that all ` relativistic ' terms involved , of the type @xmath382 ( where @xmath383 is any of the four wavespeeds involved ) for @xmath384 , are replaced for @xmath385 by @xmath386 , which he rightfully identified as a ` radiation condition ' without however giving any further explanation .
in fact , as remarked by pellegrini @xcite ( see sec . 7 and appendix a in that reference )
, the following complex identity valid for the principal determination of the square root , namely , @xmath387 for @xmath388 implies that , for @xmath368 , @xmath389 ^ 2}=-{\mathrm{i}}{\mathop{\text{sign}}}(v)\sqrt{(v+{\mathrm{i}}\epsilon)^2/c^2 - 1}.\end{aligned}\ ] ] we immediately deduce that , with @xmath390 the heaviside unit - step function , @xmath391 thus , in the isotropic case , carrying out the analytic continuation is equivalent to implementing adams s radiation - condition prescription . in the anisotropic case where ,
employing stroh s formalism , the functions involved must ultimately be computed numerically ( except in high - symmetry cases ) , the analytic continuation generalizes adams s prescription in an ` automatic ' way , alleviating the need to _ actually know _ these functions in closed analytical form .
second , we derived an explicit expression for the distortion field of an eshelby dislocation in an anisotropic medium , eq .
( [ eq : betaflat ] ) , from which the stress field immediately follows .
the expression reduces to the classical result in the volterra limit and in the subsonic regime .
however , it was extended to supersonic velocities by virtue of the above analytic continuation .
it should be noted , however , that the occurrence of the sign @xmath392 in eq .
( [ eq : betavolt ] ) is non - trivial . in the volterra limit , this allowed us to derive analytically the mach cone structure .
keeping instead the core width finite , we could effectively evaluate numerically the mach cones from eq .
( [ eq : betaflat ] ) , which was illustrated by full - field plots . in this respect ,
the present work can be seen as a continuation of our previous efforts relative to isotropic media @xcite .
third , payton s ` backward ' mach cones were given an explanation , and a simple criterion was given to determine from slowness surfaces the velocity range in which they show up .
the range is simply the one for which waves have a normal to the slowness surface ( proportional to the group velocity ) that points towards the abscissa axis .
there exists a well - known close connection @xcite between the problem of a uniformly - moving dislocation in an anisotropic media , and the theory of surface waves and of wave reflection at interfaces in anisotropic media ( see , e.g. , @xcite for recent reviews ) . in the latter context
, it is remarked that waves with such normals have been interpreted as incident waves onto the interface @xcite .
our analysis of ` backward ' cones shows that such waves can also be radiated away _ from _ the glide plane ( but the side opposite as the usual one ) , thus offering a new perspective on their practical significance .
the very natural question as to whether fully - developed ` backward ' march cones could effectively be observed in atomistic simulations or in finite - element calculations is an issue beyond the scope of this paper . to answer
it would require at least investigating the stability of such steady motions @xcite .
this problem is connected with the determination of the width of the eshelby dislocation as a function of the velocity @xcite , and will be examined elsewhere @xcite .
the author thanks a. vattr for discussions about the stroh formalism in the preliminary steps .
some useful distributions employed in the paper are briefly recalled .
first , the well - known sokhotski - plemelj formula is @xmath393 taking its derivative with respect to @xmath394 gives @xmath395 which can be generalized to any number of successive differentiations ( e.g. , @xcite , p. 94 ) .
the interesting identity ( [ eq : curious ] ) is proven as follows .
introduce the integral @xmath396 where @xmath89 is the angle between @xmath72 and @xmath100 . by ( [ eq : dsk ] )
, the limit of @xmath397 as @xmath268 is equal to the integral in the left - hand side of eq .
( [ eq : curious ] ) . then @xmath398 the real part of the integral vanishes by symmetry , so that @xmath399 the fraction in this result goes to @xmath102 as @xmath268 @xcite , which proves identity ( [ eq : curious ] ) .
we explain the computation of the @xmath174 component of integral ( [ eq : intflat1 ] ) .
the other angular integrals of the paper are obtained by the same method .
we thus need to compute @xmath400 where @xmath401 stands for @xmath402 , @xmath403 , and @xmath404 is a complex angle .
then , the imaginary parts @xmath405 and @xmath406 have same signs . introducing the complex variable @xmath407 , integral ( [ eq : ivmint ] )
is transformed into a contour integral over @xmath408 in the complex plane , on the unit circle @xmath409 defined by @xmath410 .
this change of variables entails where , invoking cauchy s theorem , the integral is computed by summing up residues @xmath412 of @xmath413 at its poles .
the first sum in ( [ eq : sumres ] ) is over the residues at those poles @xmath414 of @xmath413 that are enclosed within the contour @xmath409 ( i.e. , such that @xmath415 ) .
the rightmost one , in which we have introduced weights @xmath416 , is over the residues at _ all _ poles , provided that we take @xmath417 if @xmath415 and @xmath418
if @xmath419 . in the intermediate case @xmath420 where the pole lies _ on _ the contour @xmath409 , the principal - value prescription is invoked to handle the singularity . as discussed shortly , this simply amounts to taking @xmath421 . the function @xmath422 has seven poles @xmath414 , @xmath423 with associated residues @xmath412 . with @xmath424
, they read @xmath425}{2(\varepsilon^2 - 1 + 2{\mathrm{i}}\varepsilon s_y \cos\gamma)[\cos(\gamma-\psi)-{\mathrm{i}}s_y\varepsilon\cos\psi]}}},\\ { { \displaystyle z_{5,6}=\pm\sqrt{\frac{1+\varepsilon^2 + 2\varepsilon|\sin\gamma|}{{\mathrm{e}}^{-2{\mathrm{i}}\gamma}+\varepsilon^2 } } } } & { { \displaystyle r_{5,6}=\frac{\sin\gamma[\sin(\gamma-\psi)-{\mathrm{i}}s_y \varepsilon\sin\psi]}{2(\varepsilon^2 - 1 - 2{\mathrm{i}}\varepsilon s_y \cos\gamma)[\cos(\gamma-\psi)+{\mathrm{i}}\varepsilon s_y\cos\psi]}}}. \end{array}\end{aligned}\ ] ] in the expressions for the poles @xmath426 , we have for convenience correlated with @xmath427 the alternate sign under the square roots by introducing the absolute value of @xmath428 , so that @xmath429 ( poles inside the contour ) and @xmath430 ( poles outside the contour ) . as a result , @xmath431 always intervenes within a group @xmath432 in the residues .
the latter inequalities are saturated if and only if @xmath433 or @xmath434 , namely , on the the glide path @xmath435 in cartesian coordinates . in the generic case @xmath436 , we have @xmath437 and @xmath438 .
when @xmath439 the poles in the pairs @xmath440 and @xmath441 , and @xmath442 and @xmath443 , coalesce , pinching the contour : the integral develops a so - called _ pinch singularity _ , which sometimes leads to contour integrals being divergent . however , the singularity is inessential in the present case since the residues @xmath444 vanish in the limit .
moreover , @xmath445 , and if @xmath446 , then @xmath447 so that @xmath448 , whereas @xmath449 in the opposite situation .
it is intuitively obvious that the intermediate singular case @xmath450 ( for which both poles @xmath451 lie on the contour , and the contour integral is ill - defined ) must be handled by taking @xmath452 , i.e. , by averaging the results of both previous cases .
this amounts to taking a principal - value prescription . expressing the weighted sum of residues ( [ eq : sumres ] ) in terms of @xmath453 by means of lengthy simplifications involving usual trigonometric identities ,
we compute separately the cases @xmath446 ( only poles @xmath454 contribute ) and @xmath455 ( only poles @xmath456 contribute ) . both results are encoded into the following single expression , where @xmath457 with the convention that @xmath458 : @xmath459.\end{aligned}\ ] ] multiplying by @xmath460 as in eq .
( [ eq : intflat1 ] ) eventually yields the @xmath174 component in that equation .
calculations as above are eased by the use of an algebraic computational toolbox .
ruestes , e.m .
bringa , r.e .
rudd , b.a .
remington , t.p .
remington , m.a .
meyers , probing the character of ultra - fast dislocations , scientific reports * 5 * ( 2015 ) 16892 .
http://dx.doi.org/10.1038/srep16892[doi:10.1038/srep16892 ] .
v. nosenko , g.e .
morfill , p. rosakis , direct experimental measurement of the speed - stress relation for dislocations in a plasma crystal , phys .
106 ( 2011 ) 155002 .
http://dx.doi.org/10.1103/physrevlett.106.155002[doi:10.1103/physrevlett.106.155002 ] .
m. valle , e.m .
dunham , observation of far - field mach waves generated by the 2001 kokoxili supershear earthquake , geophys .
lett . 39 ( 2012 ) , l05311 . http://dx.doi.org/10.1029/2011gl050725[doi:10.1029/2011gl050725 ] .
j. weertman , stress dependence on the velocity of a dislocation moving on a viscously damped slip plane , in a.s . argon ( ed . ) , physics of strength and plasticity , m.i.t . press , cambridge ,
ma , 1969 , pp .
7583 .
pellegrini , equation of motion and subsonic - transonic transitions of rectilinear edge dislocations : a collective - variable approach , phys .
b 90 ( 2014 ) 054120
. http://dx.doi.org/10.1103/physrevb.90.054120[doi:10.1103/physrevb.90.054120 ] .
pellegrini , reply to `` comment on ` dynamic peierls - nabarro equations for elastically isotropic crystals ' '' .
b 83 ( 2011 ) 056102
. http://dx.doi.org/10.1103/physrevb.83.056102[doi:10.1103/physrevb.83.056102 ] .
pellegrini , screw and edge dislocations with time - dependent core width : from dynamical core equations to an equation of motion , j. mech .
solids 60 ( 2012 ) 227249 .
http://dx.doi.org/10.1103/physrevb.81.024101[doi:10.1103/physrevb.81.024101 ] .
pellegrini , m. lazar , on the gradient of the green tensor in two - dimensional elastodynamic problems , and related integrals : distributional approach and regularization , with application to non - uniformly moving sources , wave motion 57 ( 2015 ) 4463 .
http://dx.doi.org/10.1016/j.wavemoti.2015.03.004[doi:10.1016/j.wavemoti.2015.03.004 ] .
m. lazar , y .- p .
pellegrini , distributional and regularized radiation fields of non - uniformly moving straight dislocations , and elastodynamic tamm problem , j. mech .
( submitted 2015 ; in press ) . http://dx.doi.org/10.1016/j.jmps.2016.07.011[doi:10.1016/j.jmps.2016.07.011 ]
payton , steady state stresses induced in a transversely isotropic elastic solid by a moving dislocation , z. angew .
46 ( 1995 ) 282288 .
http://dx.doi.org/10.1007/bf00944758[doi:10.1007/bf00944758 ] .
f. kroupa , short range interaction between dislocations , key engineering materials 377382 ( 1995 ) 9798 ( trans .
switzerland ) .
http://dx.doi.org/10.4028/www.scientific.net/kem.97-98.377[doi:10.4028/www.scientific.net/kem.97-98.377 ] .
d. m. barnett , j. lothe , k. nishioka , r.j .
asaro , elastic surface waves in anisotropic crystals : a simplified method for calculating rayleigh velocities using dislocation theory , j. phys .
f : metal .
3 ( 1973 ) 10831096 .
http://stacks.iop.org/0305-4608/3/i=6/a=001 .
barnett , bulk , surface , and interfacial waves in anisotropic linear elastic solids , int .
j. solids struct .
37 ( 2000 ) 4554 .
http://dx.doi.org/10.1016/s0020-7683(99)00076-1[doi:10.1016/s0020-7683(99)00076-1 ] .
bacon , d.m .
barnett , r.o .
scattergood , anisotropic continuum theory of lattice defects , prog .
( 1979 ) 51262 .
http://dx.doi.org/10.1016/0079-6425(80)90007-9[doi:10.1016/0079-6425(80)90007-9 ] .
k. maln , stability and some characteristics of uniformly moving dislocations , in : j.a .
simmons , r. de wit , r. bullough ( eds . ) , fundamental aspects of dislocation theory , vol .
1 , nat . bur .
( u.s . ) .
publ . 317 , 1970 , pp .
wang , j.d .
achenbach , elastodynamic fundamental solutions for anisotropic solids , geophys .
j. int . 188
( 1994 ) 384392 .
http://dx.doi.org/10.1111/j.1365-246x.1994.tb03970.x[doi:10.1111/j.1365-246x.1994.tb03970.x ] .
budreck , an eigenfunction expansion of the elastic wave green s function for anisotropic media , q. j. mech .
( 1993 ) 126 .
http://dx.doi.org/10.1093/qjmam/46.1.1[doi:10.1093/qjmam/46.1.1 ] .
zeh , the physical basis of the arrow of time , second ed . , springer - verlag , new york , 1992 .
tewary , computationally efficient representations for elastostatic and elastodynamic green s functions for anisotropic solids , phys . rev .
b 51 ( 1995 ) 1569515702
. http://dx.doi.org/10.1103/physrevb.51.15695[doi:10.1103/physrevb.51.15695 ] .
liu , k. y. lam , two - dimensional time - harmonic elastodynamic green s functions for anisotropic media , int . j. engng .
34 ( 1996 ) 13271338 .
http://dx.doi.org/10.1016/0020-7225(96)00040-7[doi:10.1016/0020-7225(96)00040-7 ] .
wu , extension of stroh s formalism to self - similar problems in two - dimensional elastodynamics , proc .
a 456 ( 2000 ) 869890 .
http://dx.doi.org/10.1098/rspa.2000.0540[doi:10.1098/rspa.2000.0540 ] .
p. chadwick and g.d .
smith , foundations of the theory of surface waves in anisotropic elastic materials , adv .
17 ( 1977 ) 303376 .
http://dx.doi.org/10.1016/s0065-2156(08)70223-0[doi:10.1016/s0065-2156(08)70223-0 ] .
x. markenscoff and luqun ni , the transient motion of a dislocation with a ramp - like core , j. mech .
solids 49 ( 2001 ) 16031619 .
http://dx.doi.org/10.1016/s0022-5096(00)00062-4[doi:10.1016/s0022-5096(00)00062-4 ] .
a. spielmannov , a. machov , p. hora , transonic twins in 3d bcc iron crystal .
48 ( 2010 ) 296302 .
http://dx.doi.org/10.1016/j.commatsci.2010.01.010[doi:10.1016/j.commatsci.2010.01.010 ] .
m. j. lighthill , studies on magneto - hydrodynamic waves and other anisotropic wave motions , phil .
london a 252 ( 1960 ) 397430 .
http://dx.doi.org/10.1098/rsta.1960.0010[doi:10.1098/rsta.1960.0010 ] .
freund , wave motion in an elastic solid due to a nonuniformly moving line load , quart .
( 1972 ) 271281 .
http://dx.doi.org/http://www.jstor.org/stable/43636265[doi:http://www.jstor.org/stable/43636265 ] .
k. kaouri , d.j .
allwright , c.j .
chapman , j.r .
ockendon , singularities of wavefields and sonic boom , wave motion 45 ( 2008 ) 217237 .
http://dx.doi.org/10.1016/j.wavemoti.2007.06.003[doi:10.1016/j.wavemoti.2007.06.003 ] .
x. markenscoff , surong huang , analysis for a screw dislocation accelerating through the shear - wave speed barrier , j. mech .
solids 56 ( 2008 ) 22252239 .
http://dx.doi.org/10.1016/j.jmps.2008.01.005[doi:10.1016/j.jmps.2008.01.005 ] .
alshits , j. lothe , comments on the relation between surface wave theory and the theory of reflection , wave motion 3 ( 1981 ) 297310 .
http://dx.doi.org/10.1016/0165-2125(81)90023-8[doi:10.1016/0165-2125(81)90023-8 ] .
j. lothe , v.i .
alshits , surface waves , limiting waves and exceptional waves : barnett s role in the development of the theory .
mech . solids . 14 ( 2009 ) 1637 .
http://dx.doi.org/10.1177/1081286508092600[doi:10.1177/1081286508092600 ] .
n. favretto - cristini , d. komatitsch , j.m .
carcione , f. cavallini , elastic surface waves in crystals .
part 1 : review of the physics , ultrasonics 31 ( 2011 ) 653660 .
http://dx.doi.org/10.1016/j.ultras.2011.02.007[doi:10.1016/j.ultras.2011.02.007 ] . | in this work , stroh s formalism is endowed with causal properties on the basis of an analysis of the radiation condition in the green tensor of the elastodynamic wave equation .
the modified formalism is applied to dislocations moving uniformly in an anisotropic medium . in practice , accounting for causality amounts to a simple analytic continuation procedure whereby to the dislocation velocity is added an infinitesimal positive imaginary part .
this device allows for a straightforward computation of velocity - dependent field expressions that are valid whatever the dislocation velocity including supersonic regimes without needing to consider subsonic and supersonic cases separately . as an illustration ,
the distortion field of a somigliana dislocation of the peierls - nabarro - eshelby - type with finite - width core is computed analytically , starting from the green tensor of elastodynamics . to obtain the result in the form of a single compact expression , use of the modified stroh formalism
requires splitting the green function into its reactive and radiative parts . in supersonic regimes ,
the solution obtained displays mach cones , which are supported by dirac measures in the volterra limit . from these results , an explanation of payton s ` backward ' mach cones [ r. g. payton , z. angew . math . phys . *
46 * , 282288 ( 1995 ) ] is given in terms of slowness surfaces , and a simple criterion for their existence is derived .
the findings are illustrated by full - field calculations from analytical formulas for a dislocation of finite width in iron , and by huygens - type geometric constructions of mach cones from ray surfaces . a stroh formalism , b moving dislocations , c mach cones * |
quantum chromodynamics ( qcd ) is the only quantum field theory of the standard model whose _ collective _ behaviour ( phase diagram and phase transitions ) is accessible to study in the laboratory .
high - energy heavy - ion ( aa ) collisions offer the only experimental means known so far to concentrate a significant amount of energy ( @xmath0(1,10 tev ) at rhic , lhc ) in a `` large '' volume ( @xmath1 @xmath2 ) at thermalization times of @xmath3 1 fm/@xmath4 ) .
the study of the many - body dynamics of high - density qcd covers a vast range of fundamental physics problems : * * deconfinement and chiral symmetry restoration * : lattice qcd calculations predict a new form of matter at energies densities above @xmath5 1 gev/@xmath2 consisting of an extended volume of deconfined and bare - mass quarks and gluons : the quark gluon plasma ( qgp ) @xcite .
the scrutiny of this new state of matter ( equation - of - state , order of the phase transition , ... ) promises to shed light on fundamental questions of the strong interaction such as the nature of confinement , the mechanism of mass generation ( chiral symmetry breaking , structure of the qcd vacuum ) and hadronization , which still evade a thorough theoretical description due to their highly non - perturbative nature @xcite . * * early universe cosmology * : the quark - hadron phase transition took place some 10 @xmath6s after the big - bang and was the most important event taking place in the universe between the electro - weak ( or susy ) transition ( @xmath7 s ) and big bang nucleosynthesis ( @xmath8 200 s ) . depending on the order of the qcd phase transition , several cosmological implications such as the formation of strangelets and cold dark - matter ( wimp ) clumps or baryon fluctuations leading to inhomogeneous nucleosynthesis , have been postulated @xcite . * * parton structure and evolution at small-@xmath9 * : at high energies , hadrons consist of a very dense system of gluons with small ( bjorken ) parton fractional momenta @xmath10 . at low-@xmath9 ,
the probability to emit an extra gluon is large @xmath11 and non - linear @xmath12 processes will eventually dominate the parton evolution in the hadronic wave functions . whereas hera results indicate that for @xmath1310@xmath14 , the parton evolution with @xmath15 ( or @xmath16 ) is described by the usual dglap ( or bfkl ) equations , at lower values of @xmath9 and around a saturation momentum of @xmath17 2 gev@xmath18 , such a configuration is theoretically described in terms of the `` colour glass condensate '' ( cgc ) picture @xcite .
since the nonlinear growth of the gluon density depends on the transverse size of the system ( i.e. @xmath19 , where @xmath20 is the number of nucleons in the nucleus ) , the effects of gluon saturation are expected to set in earlier for ultrarelativistic heavy nuclei than free nucleons . *
* gauge / string duality * : theoretical calculations based on the anti - de - sitter / conformal - field - theory ( ads / cft ) correspondence permit to obtain results in strongly coupled ( @xmath21 ) su(@xmath22 ) gauge theories in terms of a weakly - coupled dual gravity theory @xcite .
recent applications of this formalism for qcd - like @xmath23 super yang - mills theories have allowed to compute transport properties of experimental relevance , - such as the qgp viscosity @xcite , the `` jet quenching '' parameter @xmath24 @xcite , or the heavy - quark diffusion coefficient @xcite - , from black hole thermodynamics calculations .
such results provide valuable insights on dynamical properties of non - perturbative qcd that can not be directly treated by lattice methods .
in this overview , we present a _
selection _ of experimental results ( mostly from the comprehensive reviews of the 4 rhic experiments @xcite ) from auau , dau and pp collisions up to a maximum center - of - mass energy of @xmath25 = 200 gev .
direct information on the thermodynamical and transport properties of the strongly interacting medium produced in aa collisions is commonly obtained by comparing the results for a given observable @xmath26 to those measured in p(d)a ( `` cold qcd matter '' ) and in pp ( `` qcd vacuum '' ) collisions as a function of center - of - mass energy , @xmath27 , rapidity @xmath28 , reaction centrality ( impact parameter @xmath29 ) , and particle type ( mass ) : @xmath30 the observed _ enhancements _ ( e.g. in photon or baryon yields , or soft hadron slopes ) and/or _ suppressions _ ( e.g. in total multiplicities , high-@xmath27 leading hadrons , or quarkonia yields ) in the @xmath31 ratios
can be directly related to the properties of the produced qcd matter after accounting for a realistic modeling of the space - time evolution of the collision process and stop self - interacting collectively at freeze - out times @xmath32 15 fm/@xmath4 . ] .
the bulk hadron multiplicities measured at mid - rapidity in central auau at @xmath25 = 200 gev are @xmath35 700 , comparatively lower than the @xmath35 1000 expectations @xcite of `` minijet '' dominated scenarios , soft regge models ( without accounting for strong shadowing effects ) , or extrapolations from an incoherent sum of proton - proton collisions ( fig .
[ fig : dndeta ] , left ) . on the other hand ,
approaches @xcite based on gluon saturation @xcite , which take into account a reduced _ initial _ number of scattering centers in the nuclear parton distribution functions , @xmath36 , agree well with experimental data . in those cgc calculations ,
the final hadron multiplicities are assumed to be simply related to the initial number of released partons ( local parton - hadron duality ) which are depleted in the initial state compared to proton - proton collisions due to non - linear gluon - gluon fusion effects . in auau at @xmath25 = 200 gev @xcite .
right : energy and centrality dependences ( in terms of the number of nucleons participating in the collision , @xmath37 ) of @xmath38 ( normalized by @xmath37 ) : phobos auau data @xcite versus the predictions of the saturation approach @xcite . ,
title="fig:",width=283,height=264 ] in auau at @xmath25 = 200 gev @xcite .
right : energy and centrality dependences ( in terms of the number of nucleons participating in the collision , @xmath37 ) of @xmath38 ( normalized by @xmath37 ) : phobos auau data @xcite versus the predictions of the saturation approach @xcite . , title="fig:",width=317 ] -0.7 cm the good reproduction of the bulk aa hadron multiplicities ( including its centrality and center - of - mass energy dependences , fig .
[ fig : dndeta ] right ) has been one of the supporting arguments in favour of the existence of non - linear qcd effects in high - energy nuclear collisions .
in addition , the brahms observation of suppressed yields of moderately high-@xmath27 hadrons in dau at forward rapidities ( @xmath39 3.2 ) @xcite , as well as the `` geometrical scaling''-like behaviour of the nuclear pdfs observed for @xmath400.017 @xcite , are also consistent with gluon saturation expectations .
it is worth noting , however , that both results are in a kinematic range with relatively low momentum scales , @xmath0(1 gev ) , where non - perturbative effects can blur a simple interpretation based on partonic degrees of freedom alone . indeed , at rhic ( and hera )
energies the saturation momentum , - the scale at which non - linear effects become important and start to saturate the parton densities - , is in the transition between the soft and hard regimes ( @xmath41 2 gev@xmath18 ) . at lhc
the relevance of low-@xmath9 qcd effects in hadronic collisions will be certainly enhanced due to the increased ( i ) center - of - mass energy @xmath25 , ( ii ) nuclear radius @xmath42 , and ( iii ) rapidity @xmath28 of the produced partons .
indeed , at lhc energies not only the relevant bjorken @xmath9 values will be 3070 times lower than at rhic : @xmath43 at central ( forward ) rapidities for processes with a hard scale @xmath4410 gev , but also the saturation momentum , @xmath45 5 10 gev@xmath18 @xcite , will be in the perturbative range .
the bulk of hadron production ( @xmath46 2 gev/@xmath4 ) in auau at rhic shows strong collective effects known as radial and elliptic flows .
on the one hand , the measured single hadron @xmath27 spectra have an inverse slope parameter larger than that measured in pp collisions , increasing with reaction centrality and hadron mass as expected if collective expansion effects blue - shift the hadron spectra ( fig .
[ fig : hydro ] , left ) .
phenomenological fits to `` blast wave '' models yield transverse flow velocities @xmath47 0.6 @xcite . on the other hand ,
the azimuthal distribution @xmath48 of hadrons emitted w.r.t .
the reaction plane show a strong harmonic modulation with a preferential `` in - plane '' emission in non - central collisions .
such an azimuthal flow pattern is a truly collective effect ( absent in pp collisions ) consistent with an efficient translation of the initial @xmath9-space anisotropy in non - central reactions ( with an almond - shape overlap zone ) into a final `` elliptical '' asymmetry in @xmath49-space .
the amount of elliptic flow is quantified via the second fourier coefficient @xmath50 of the @xmath48 distribution relative to the reaction plane .
the large @xmath51 0.2 measured in the data ( fig .
[ fig : hydro ] , right ) indicates a strong degree of collectivity ( pressure gradients ) developing in the first instants of the collision .
indeed , elliptic flow develops in the initial phase of the reaction and quickly self - quenches beyond @xmath32 5 fm/@xmath4 as the original spatial eccentricity disappears .
this is confirmed by the observation that not only light hadrons but charm quarks ( indirectly measured via the semileptonic decays of d mesons into @xmath52 ) show a @xmath53 signal as large as 10% @xcite clearly consistent with strong collective correlations during the _ partonic _ phase of the reaction .
5 gev/@xmath4 in 0 - 10% most central auau collisions at @xmath25 = 200 gev , compared to hydrodynamics(+pqcd ) calculations @xcite .
right : measured elliptic flow parameter @xmath54 for a variety of hadrons @xcite compared to hydrodynamic predictions @xcite.,title="fig:",width=264 ] 5 gev/@xmath4 in 0 - 10% most central auau collisions at @xmath25 = 200 gev , compared to hydrodynamics(+pqcd ) calculations @xcite .
right : measured elliptic flow parameter @xmath54 for a variety of hadrons @xcite compared to hydrodynamic predictions @xcite.,title="fig:",width=302 ] -0.7 cm interestingly , the strong @xmath53 seen in the data is inconsistent with the much lower values , @xmath55 6% , expected for purely hadronic matter @xcite as well as for a partonic system cascading with perturbative cross - sections ( @xmath56 3 mb ) @xcite .
the magnitude , @xmath27 and mass dependences of the radial and elliptic flows below @xmath57 2 gev/@xmath4 are , however , remarkably well described by _
ideal _ hydrodynamics models whose space - time evolution starts with a qgp equation - of - state ( eos ) with initial @xmath58 30 gev/@xmath2 at very short thermalization times @xmath3 0.6 fm/@xmath4 @xcite ( fig . [
fig : hydro ] ) .
the fast ( local ) thermalization times , the robust collective flow generated in the first instants of the reaction , and the excellent agreement of the data with ideal hydrodynamics models which assume a fluid evolution with zero viscosity , have been presented @xcite as evidences that the qgp formed at rhic is a strongly coupled liquid ( characterized by coulomb coupling parameter @xmath59 1 ) rather than a weakly interacting gas of partons .
estimates of the maximum amount of viscosity allowed by the @xmath54 data @xcite give a value close to the conjectured universal lower bound for the dimensionless viscosity / entropy ratio , @xmath60 , obtained from ads / cft calculations @xcite .
this result would make the qgp the most perfect fluid ever observed .
the measurement of the differential elliptic flow properties in aa collisions at lhc will be of primary importance to confirm such an interpretation and search for a possible transition from a hotter weakly interacting qgp to the liquid - like state found at rhic @xcite .
among the most exciting results of the rhic physics programme is the observed strong suppression of high-@xmath27 leading hadron spectra in central aa @xcite consistent with the predicted energy loss of the parent light quarks and gluons traversing the dense colored medium ( `` jet quenching '' ) @xcite . above @xmath57 5 gev/@xmath4 , @xmath63 , @xmath64 , and
inclusive charged hadrons show all a common factor of @xmath655 suppression compared to an incoherent superposition of pp collisions ( fig .
[ fig : r_aa_rhic_200 ] , left ) @xcite .
the @xmath66 = 1 perturbative expectation which holds for other hard probes such as `` color blind '' direct photons and for high-@xmath27 hadrons in dau reactions ( where no final - state dense and hot system is produced ) @xcite , is badly broken ( @xmath67 0.2 ) in central auau collisions . for different particles measured in central auau at @xmath25 = 200 gev
compared to parton energy loss model predictions . left
: direct @xmath68 , @xmath63 , and @xmath64 @xcite .
right : `` non - photonic '' @xmath52 , mainly from d , b meson decays @xcite.,title="fig:",width=321,height=207 ] for different particles measured in central auau at @xmath25 = 200 gev compared to parton energy loss model predictions .
left : direct @xmath68 , @xmath63 , and @xmath64 @xcite .
right : `` non - photonic '' @xmath52 , mainly from d , b meson decays @xcite.,title="fig:",height=198 ] -0.7 cm most of the empirical properties of the quenching factor for light - flavor hadrons ( magnitude , @xmath27- , centrality- , @xmath25- dependences of the suppression ) @xcite are in quantitative agreement with the predictions of non - abelian parton energy loss models which assume that the parent parton loses energy by gluonstrahlung while traversing a medium with a large color density .
very large initial gluon rapidity densities , @xmath69 1000 @xcite , or equivalently , transport coefficients characterizes the squared average @xmath70 transfer from the medium to the hard parton per unit distance @xcite . ] , @xmath71 14 gev@xmath18/fm @xcite , are needed to explain the amount of hadron suppression . however , the fact that the quenching factor for high-@xmath27 electrons from semi - leptonic @xmath72 and @xmath73 decays is as suppressed as the light hadrons in central auau ( fig .
[ fig : r_aa_rhic_200 ] , right ) @xcite is in apparent conflict with the robust @xmath74 prediction of radiative energy loss models . in order to reproduce the high @xmath27 open charm / bottom suppression ,
jet quenching models require either initial gluon densities ( @xmath69 3000 ) inconsistent with those needed to describe the quenched light hadron spectra @xcite , or a smaller relative contribution of @xmath73 relative to @xmath72 mesons than theoretically expected in the measured decay electron @xmath27 range @xcite .
this discrepancy may point to an additional contribution from elastic ( i.e. non - radiative ) energy loss @xcite for heavy - quarks @xcite which was considered negligible so far .
the unique possibility at lhc of fully reconstructing @xmath75 jets will be very valuable to clarify the response of strongly interacting matter to fast _ heavy _ quarks , and provide detailed additional information on the transport properties of qcd matter @xcite .
a second striking observation of the jet - quenching phenomena at rhic is the strongly modified azimuthal dijet correlations compared to baseline pp results . due to the difficulties of full jet reconstruction in aa at rhic ,
jet - like correlations are measured on a statistical basis by selecting the highest @xmath27 _ trigger _ hadron of the event and measuring the azimuthal ( @xmath78 ) and rapidity ( @xmath79 ) distributions of _ associated _ hadrons ( @xmath80 ) : @xmath81 . in pp collisions
, a dijet signal appears as two distinct back - to - back gaussian peaks at @xmath82 0 ( near - side ) and at @xmath83 ( away - side ) . at variance with this standard dijet topology in the qcd vacuum , the away - side dihadron azimuthal correlations @xmath84 in central auau collisions shows a `` dip '' with a `` double peak '' structure at @xmath85 1.1 ( fig .
[ fig : dndphi ] ) for semihard associated hadrons ( @xmath86 = 1 2.5 gev/@xmath4 ) @xcite .
= 1 2.5 gev/@xmath4 ) with respect to a trigger hadron measured at rhic . left : phenix data in central ( top ) and peripheral ( bottom ) auau @xcite .
right : star data in central auau and dau collisions @xcite.,title="fig:",width=283]0.5 cm = 1 2.5 gev/@xmath4 ) with respect to a trigger hadron measured at rhic . left : phenix data in central ( top ) and peripheral ( bottom ) auau @xcite .
right : star data in central auau and dau collisions @xcite.,title="fig:",width=294,height=257 ] -0.7 cm such a non - gaussian ( `` volcano''-like ) profile in the away - side hemisphere has been interpreted as due to the preferential emission of energy from the quenched parton at a finite angle with respect to the jet axis .
such a conical - like pattern can appear if the lost energy excites a collective mode of the medium and generates a wake of lower energy gluons with mach- @xcite or erenkov - like @xcite angular emissions .
in the first case , the _ speed of sound _ of an ultrarelativistic fluid is a simple proportionality constant relating its pressure and energy density : @xmath87 .
] of the traversed matter , @xmath88 , can be determined from the characteristic supersonic angle of the emitted secondaries : @xmath89 , where @xmath90 is the mach shock wave angle . the resulting preferential emission of secondary partons from the plasma measured at a _ fixed _
angle @xmath91 , yields an average value of the speed sound @xmath92 0.45 , larger than that of a hadron gas ( @xmath93 0.35 ) @xcite , and not far from that of an ideal qgp ( @xmath94 ) .
a selection of experimental data from central auau collisions at rhic energies ( @xmath25 = 200 gev ) has been presented providing direct information on fundamental thermodynamical and transport properties of high - density qcd matter .
four notable experimental results have been discussed : ( i ) the reduced total hadron multiplicities consistent with gluon saturation in the initial nuclear parton distribution functions , ( ii ) the strong transverse and elliptic differential flows in the bulk hadron spectra indicative of a high degree of collectivity and very low viscosities during the first instants of the reaction , ( iii ) the factor @xmath655 suppression of high @xmath27 leading hadrons reproduced by parton energy loss calculations for a medium with very large initial gluon densities ( @xmath69 1000 ) and transport coefficient ( @xmath71 14 gev@xmath18/fm ) , and ( ii ) the non - gaussian shape of the azimuthal distributions of secondary hadrons in the away - side hemisphere of high-@xmath27 trigger hadrons attributed to mach conical flow caused by the propagation of a supersonic parton through the dense system .
other interesting probes of quark - gluon matter ( photons , quarkonia , ... ) , not covered here by lack of space , will be discussed in a coming publication @xcite .
nucleus - nucleus collisions at lhc energies will undoubtedly contribute to expand the knowledge of many - body qcd at extreme conditions of temperature , density and low-@xmath9 shedding light on a vast ramification of fundamental physics problems .
+ * acknowledgments * + this work is supported by the 6th eu framework programme contract meif - ct-2005 - 025073 .
see e.g. f. karsch , lect .
notes phys . 583 ( 2002 )
m. jaffe and e. witten , `` quantum yang - mills theory '' , clay maths inst .
millennium prize ( 2000 ) , ` http://www.claymath.org/millennium/yang-mills_theory ` ; and d. gross , in `` ten problems in fundamental physics '' ( 2000 ) , ` http://feynman.physics.lsa.umich.edu/strings2000/millennium.html ` .
d. j. schwarz , annalen phys .
12 ( 2003 ) 220.see e.g. e. iancu and r. venugopalan , in `` qgp .
vol 3 '' eds : r.c .
hwa and x.n .
wang , world scientific , singapore , hep - ph/0303204 .
j. m. maldacena , adv .
theor . math .
. 2 ( 1998 ) 231 [ int .
j. theor .
phys . 38 ( 1999 ) 1113 ] ; e. witten , adv .
phys . 2 ( 1998 ) 505.p .
kovtun , d. t. son and a. o. starinets , phys .
lett . * 94 * ( 2005 ) 111601.h .
liu , k. rajagopal and u. a. wiedemann , hep - ph/0605178 .
j. casalderrey - solana and d. teaney , hep - ph/0605199 .
[ phenix ] , nucl .
a757 ( 2005 ) 184 .
j. adams _ et al . _ [ star ] , nucl . phys .
a757 ( 2005 ) 102 . b. b. back _ et al .
_ [ phobos ] , nucl .
a757 ( 2005 ) 28 .
i. arsene _
[ brahms ] , nucl .
a757 ( 2005 ) 1 . k. j. eskola , nucl .
a * 698 * ( 2002 ) 78 .
d. kharzeev and m. nardi , phys .
b * 507 * ( 2001 ) 121 ; d. kharzeev , e. levin and m. nardi , nucl . phys . a * 747 * ( 2005 ) 609 .
n. armesto , c. a. salgado and u. a. wiedemann , phys .
* 94 * ( 2005 ) 022002 .
d. denterria and d. peressounko , eur .
j. c * 46 * ( 2006 ) 451 y. akiba [ phenix ] , nucl - ex/0510008 .
r. snellings , aip conf .
* 756 * ( 2005 ) 390 .
p. huovinen and p. v. ruuskanen , ann .
nucl . part .
sci . to appear , nucl - th/0605008.x .
l. zhu , m. bleicher and h. stoecker , phys .
c * 72 * ( 2005 ) 064911 .
d. molnar and m. gyulassy , nucl .
a * 697 * ( 2002 ) 495 [ errat .- ibid .
a * 703 * ( 2002 ) 893 ] p. f. kolb and u. heinz , in `` qgp .
vol 3 '' eds : r.c .
hwa and x.n .
wang , world scientific , singapore , nucl - th/0305084 , and refs .
therein.d .
teaney , j. lauret and e. v. shuryak , nucl - th/0110037.t .
hirano and k. tsuda , phys .
c * 66 * ( 2002 ) 054905.e .
shuryak , prog .
* 53 * ( 2004 ) 273 .
m. h. thoma , j. phys .
g * 31 * ( 2005 ) l7 .
t. hirano and m. gyulassy , nucl .
a * 769 * ( 2006 ) 71 .
a. peshier and w. cassing , phys .
* 94 * ( 2005 ) 172301 .
u. w. heinz , nucl - th/0512051 .
[ phenix ] , ( 2002 ) 022301 ; s.adler _ et al .
_ [ phenix ] , ( 2003 ) 072301 ; j.adams _ et al . _
[ star ] , ( 03 ) 172302.m .
gyulassy , m. plmer , ( 1990 ) 432 ; x.n .
wang , m. gyulassy , ( 1992 ) 1480 .
s. s. adler _ et al . _
[ phenix ] , phys .
* 96 * ( 2006 ) 202301 ; m. shimomura [ phenix ] , nucl - ex/0510023.s .
s. adler _ et al . _
[ phenix ] , ( 2003 ) 072303 ; j. adams _ et al . _ [ star ] , phys .
* 91 * ( 2003 ) 072304 .
d. denterria , ( 2005 ) 295.i .
vitev and m. gyulassy , ( 2002 ) 252301.r .
baier , y. l. dokshitzer , a. h. mueller , s. peign and d. schiff , ( 1997 ) 265 ; r. baier , d. schiff , b.g .
zakharov , ann .
nucl . part .
* 50 * ( 2000 ) 37.a .
dainese , c. loizides and g. paic , ( 2005 ) 461.n .
armesto , a. dainese , c. a. salgado and u. a. wiedemann , ( 2005 ) 054027 ; n. armesto _ et al .
_ , phys .
b * 637 * ( 2006 ) 362 .
s. s. adler _ et al . _
[ phenix ] , phys .
* 96 * ( 2006 ) 032301 b. i. abelev _ et al . _ [ star ] , nucl - ex/0607012 .
m. djordjevic , m. gyulassy and s. wicks , ( 2005 ) 112301 .
s. wicks , w. horowitz , m. djordjevic and m. gyulassy , nucl - th/0512076 .
m. g. mustafa and m. h. thoma , acta phys . hung . a * 22 * ( 2005 ) 93 .
a. peshier , hep - ph/0607275 .
s. s. adler _ et al . _
[ phenix ] , to appear in , nucl - ex/0507004.t .
peitzmann [ star ] , proceeds .
hard probes06 .
h. stoecker , , 121 ( 2005 ) ; j. casalderrey , e. shuryak , d. teaney , j. phys .
* 27 * ( 2005 ) 22.j .
ruppert and b. muller , ( 2005 ) 123.v .
koch , a. majumder and x. n. wang , phys .
* 96 * ( 2006 ) 172302 ; i. m. dremin , nucl .
phys . a * 767 * ( 2006 ) 233 .
b. mohanty and j. e. alam , phys .
c * 68 * ( 2003 ) 064903 .
d. denterria , in preparation . | a selection of experimental results in high - energy nucleus - nucleus collisions after five years of operation of the relativistic heavy - ion collider ( rhic ) is presented .
emphasis is put on measurements that provide direct information on fundamental properties of high - density qcd matter . the new experimental opportunities accessible at lhc are introduced , in particular those that may help clarify some of the current open issues at rhic . |
the continous increase of laser intensity is making possible the study of a wide range of non - linear phenomena in atoms , plasmas and solids under the action of intense electromagnetic radiation @xcite . in the last decade
these studies have been extended to semiconductor nanostructures under intense electric fields , originated by an applied ac voltage or a high - intensity infrared laser . at frequencies of the order of @xmath0 , typical of free - electron lasers , photon energies are comparable to the energy separation of the electronic levels and nanostructures couple strongly to the electromagnetic field .
in the case of semiconductor superlattices of period @xmath1 , driven by electromagnetic radiation of frequency @xmath2 , amplitude @xmath3 and polarized in a direction perpendicular to the interfaces , the i vs v curves are modified and non - linear behaviour is observed .
holthaus @xcite has predicted the collapse of minibands and the dynamical localization of electrons with the corresponding supression of the coherent tunneling current if @xmath4 is a zero of the bessel function @xmath5 , being @xmath6 . as a consequence of tucker s formula @xcite , tunneling processes in which the absorption or emission of n photons take place
are inhibited when @xmath4 is a zero of @xmath7 .
sequential photon - assisted tunneling , with differential negative resistivity , dynamical electron localization , formation of electric field domains and absolute negative conductance for low applied voltages have been observed experimentally @xcite . in the miniband transport regime
the effect of inverse bloch oscillations has been reported @xcite .
many of these phenomena have been theoretically explained @xcite and several interesting predictions have been made : oscillations of the superlattice reflection coefficient of the radiation due to dynamical electron localization @xcite , step - like behaviour of the interminiband multiphotonic absorption coefficient @xcite , spontaneous breaking of temporal symmetry with the appearance of a dc current in presence of an ac voltage @xcite and the occurrence of rabi oscillations when @xmath8 is close to the energy difference between two minibands @xcite .
a new kind of microwave frequency multiplier , an ultra - fast detector of @xmath9 radiation and a bloch laser based in the driven motion of the electrons in one miniband have been proposed @xcite .
the energy spectrum of a superlattice of period @xmath1 under the action of a magnetic field @xmath10 perpendicular to the growth direction has also been well studied @xcite . instead of bloch minibands ,
one has landau magnetic subbands , with energies @xmath11 , where @xmath12 , ( @xmath13 ) is the position of the centre of the cyclotronic orbit , @xmath14 is the momentum along the magnetic field direction and @xmath15 the conduction electron effective mass .
let @xmath16 be the magnetic length .
if the ratio @xmath17 , the first subbands are almost flat and those of higher index @xmath18 present a dispersive character . if @xmath19 , all subbands are dispersive .
the difference between these two regimes is observed in optical intraband absorption at frequencies close to that of cyclotronic resonance @xmath20 @xcite .
the dispersion of magnetic subbands makes possible tunneling in the growth direction , which occurs via transitions between consecutive subbands in the anti - crossing points @xcite .
this work considers a new , combined situation , in which a superlattice undergoes the action of a magnetic field perpendicular to its growth direction and intense electromagnetic radiation , linearly polarized in a direction perpendicular to @xmath21 .
we search for the occurrence of effects such as the collapse of magnetic subbands and electronic localization similar to those predicted for bloch minibands .
the magnetic field introduces a new characteristic frequency ( the cyclotron frequency ) in the system under consideration , and an enhanced effect of the external electromagnetic field is expected near the resonance conditions .
additionally , @xmath21 couples the motion in the directions perpendicular to it and , therefore , a similar result should be observed for waves polarized in either of these directions .
we have found that non - linear effects become important for relatively lower intensities and radiation polarized in any direction perpendicular to the magnetic field .
quasienergies and density of states in the kramers - henneberger approximation are calculated and the conditions under which collapse of magnetic subbands and supression of n - photon absorption or emission processes occur are analyzed .
let @xmath22 be the growth direction of the superlattice and @xmath23 the magnetic field direction .
the electromagnetic radiation of frequency @xmath24 and amplitude @xmath3 can be polarized either in @xmath22 or @xmath25 directions . in the effective mass approximation , the envelope wavefunction of a conduction electron can be written as @xmath26 where @xmath27 is the transverse area of the sample , and @xmath15 is the conduction electron effective mass .
we have considered that @xmath15 is constant , independent of @xmath28 .
the validity of this , commonly used , approximation has been discussed in @xcite . the time - dependent unitary transformation @xmath29 is given by @xcite : @xmath30 with the choice @xmath31 for @xmath32 , the function @xmath33 satisfies the equation @xmath34 where @xmath35\\ & & x_0=\frac{l_b^2}{\hbar}q\end{aligned}\ ] ] here @xmath36 are the fourier coefficients of the superlattice potential : @xmath37 and @xmath38 takes values @xmath39 or @xmath40 for electromagnetic waves polarized along @xmath22 or @xmath25 axis .
the effect of radiation is expressed in the `` dressed '' potential @xmath41 , which renormalizes magnetic subbands , and the superposition of harmonic terms with amplitude @xmath42 , giving raise to n - photon emission and absorption processes . the dynamical stark effect is included in eq.([choice ] ) for @xmath43 as a phase shift of the wavefunction , and is given by @xmath44 when @xmath45 , @xmath46 and standard time - dependent perturbation theory can be applied .
non - linear effects become important for @xmath47 .
if @xmath48 , all @xmath49 have the same order of magnitude and @xmath50 .
when @xmath2 is close to @xmath51 , @xmath52 is relatively small .
for example , if @xmath53 , @xmath54 and @xmath55 , then @xmath56 .
this corresponds to intensities about @xmath57 , which can be reached with a c.w .
therefore , the dynamics of recombination processes and many - body effects due to high levels of photoexcitation need not to be considered .
quasienergies @xmath58 are the eigenvalues of the operator @xmath59 @xcite and can be obtained from the poles of @xmath60 , the retarded green s function averaged over a period @xmath61 of the initial time .
the kramers - henneberger approximation ( kha ) , widely used in atoms @xcite , approximates @xmath62 by the eigenvalues of @xmath63 and corresponds to the zeroth order term of the expansion of @xmath60 in even powers of @xmath64 : @xmath65 where @xmath66 denotes the time - average operation described above .
following the arguments of @xcite , it is easy to show that the above series converges absolutely . note that @xmath67 enters the expansion in zeroth order , whereas @xmath49 ( @xmath68 ) contributes only to second and higher orders .
when @xmath48 eq.([expgreen ] ) is an expansion in powers of @xmath69 .
then , as kha neglects terms of the order of @xmath70 it is expected to give a good result if @xmath71 . since in this case @xmath72 , the dressed potential
can also be considered an small perturbation partially breaking the degeneracy of landau levels .
it is important to realize that the limit @xmath73 can be achieved not only by increasing the laser intensity ( @xmath74 ) but also by approaching the cyclotronic frequency ( @xmath75 ) . in what follows
, we will consider a particular potential : @xmath76 for which the expressions of @xmath42 have the simple form @xmath77\ ] ] in this case , magnetic subbands collapse when @xmath4 is a zero of @xmath5 , and processes involving absorption or emission of n photons are quenched at zeros of @xmath7 .
however , for an arbitrary superlattice , @xmath42 is given by eqs.([gral ] ) and different harmonics are modified in different proportions , so @xmath42 does not necessarily become zero for finite values of @xmath4 .
when @xmath78 ( @xmath73 ) and @xmath79 is fixed , @xmath49 approaches zero as @xmath80 and magnetic subbands are supressed . in this
limit multiphoton processes of many orders have a contribution of similar magnitude to the absorption and emission probabilities . to order @xmath81 the quasienergies are given by the expression @xmath82 where @xmath83 and @xmath84 are laguerre polynomials .
numerical diagonalization of @xmath63 gives the electron quasienergies @xmath85 in the kha , shown in fig.1 ( solid lines ) for @xmath86 , @xmath54 , @xmath53 , @xmath55 and @xmath87 , corresponding to laser intensities lower than @xmath88 at frequencies @xmath89 .
we have also shown the quasienergies obtained from eq.([perturb])(dashed lines ) . 1 . in the absence of radiation ( @xmath90 ) all magnetic subbands
are dispersive , in agreement with @xcite , since for the values of @xmath10 and @xmath1 considered @xmath91 .
2 . as @xmath4 increases , the magnetic subbands become flat .
3 . as @xmath4 increases , the results derived from eq.([perturb ] ) approach those calculated in kha approximation .
4 . the facts described in b. and c. do not occur monotonously , because of the oscillations of @xmath5 .
for example , these effects are stronger for @xmath92 than for @xmath93 because @xmath5 has a zero in @xmath94
one must expect the same qualitative behaviour for any superlattice , but in the general case there are no finite values of @xmath4 for which magnetic subbands collapse . note that in this case @xmath95 in fig.2 we show the density of electron states ( dos ) calculated from eq.([perturb ] ) for @xmath96 and @xmath97 .
when magnetic subbands become flat , the time averaged velocity in the @xmath25 direction is zero and electron motion becomes one - dimensional , as can be seen in the shape of the dos curve for @xmath92 . under these conditions , tunneling in the direction of growth
is expected to be quenched , as for bloch minibands .
h. s. brandi , l. davidovich , g. jalbert , b. koiller and n. zagury , in _ intense laser phenomena and related subjects .
series of optics and photonics _ , edited by i. yu .
kyian and m. yu .
ivanov , ( world scientific , singapore , 1991 ) , p. 299 . | the effect of strong laser radiation on magnetic subbands in semiconductor superlattices is investigated . due to the presence of a magnetic field perpendicular to the growth direction , non - linear effects such as band supression and electron localization
become relevant at relatively lower intensities and for any polarization perpendicular to the magnetic field .
electron quasienergies and density of states are calculated in the kramers - henneberger approximation , whose validity is discussed .
the conditions under which collapse of magnetic subbands and quenching of n - photon emission or absorption processes occur are discussed .
we conclude that at laser frequencies close to cyclotronic frequency and intensities typical of c.w .
lasers , magnetic subbands become flat , magnetotunneling is inhibited and multiphotonic processes dominate optical absorption . 2 |
a motion of point vortices ( pvs ) is one of the simple and fundamental issues in fluid dynamics @xcite .
there have been a number of numerical studies of pvs , whose motion is bounded by a circular wall .
recently , stable triangular vortex lattices have been observed in rotating bose - einstein condensates of na atoms @xcite in which the circulation of vortices is quantized with the same strength and the lattice contains over 100 vortices .
many experimental reports of superfluid vortex lattices increased in these several years show triangular patterns @xcite , which can be compared with the stability analysis of vortex lattices by tkachenko @xcite in 1966 .
conformal theory of irrotational flows and complex functions are invoked and the role of the weierstrass zeta function is crucial . this paper supplements the letter @xcite by showing the spatial distribution of pvs for various type of initial conditions .
tkachenko s work 40 years ago and recent strong interests by physicists on vortex lattices in bose - einstein condensates stimulate the author to consider the numerical study of motions of pvs with periodic boundary conditions . in order to quantify clustering of the point distribution ,
we apply point process theory used in spatial ecology .
the computed @xmath3 and @xmath2 functions judge the distribution to be clustered or not .
this method can be applied to other systems which include information of the point distribution in two or more dimensions .
the velocity field by a single periodic pv of strength @xmath4 is given by @xcite @xmath5 where @xmath6 and the overbar denotes the complex conjugate .
the corresponding streamfunction is denoted by @xmath7 the weierstrass zeta and sigma functions ( @xmath8 and @xmath9 ) contain two complex parameters @xmath10 and @xmath11 , which are half periods on the complex plane .
the second term including @xmath12 $ ] expresses the solid rotation in order to cancel the circulation on sides of the parallelogram induced by the pv . the equations of motions for pvs of strength @xmath13 , @xmath14 ,
can be described by @xmath15 for a single type of pvs with the total number @xmath16 , the index @xmath17 is @xmath18 . for positive and negative pvs with the same strength ,
the index is @xmath19 .
there are three known conserved quantities ; the hamiltonian and two components of the linear impulse @xcite .
the @xmath3 function used in spatial ecology @xcite is defined by @xmath20 where @xmath16 is the total number of points , @xmath21 is the number density , @xmath22 is the area , @xmath23 is the position of the @xmath24th points , and @xmath25 is the step function .
a function for the edge correction on the right hand side of ( [ eq4 ] ) , which usually appears in data analysis of spatial ecology , is unnecessary in the present periodic case .
the @xmath2 function is given by @xmath26 if the distribution of points is complete spatial randomness ( csr ) , then @xmath27 and @xmath28 .
if the points are clustered , @xmath29 .
the uniformly spaced ( us ) distribution leads to @xmath30 .
it is generally difficult to distinguish between csr and us distributions only if we see raw data of the point distribution .
however , the @xmath2 function gives judgment on the degree of clustering clearly .
we note that the value of @xmath31 depends on @xmath32 . for examples ,
the checkered pattern gives a positive value of @xmath31 for @xmath33 , where @xmath34 is the size of the smallest square , but @xmath35 for @xmath36 .
the system ( [ eq3 ] ) with @xmath16 up to 100 is simulated numerically by using _
mathematica _ 5.2 . the textbook by trott ( 2006 ) @xcite gives a number of numerical codes for _ mathematica _ programming , including samples of point vortices with other boundary conditions .
we first consider the following four initial conditions : ( i ) an infinite row that is a discrete model of the vortex sheet , ( ii ) pvs located randomly in checkered patterns , ( iii ) pvs located randomly in the 10 @xmath37 10 subsquares , and ( iv ) csr in the unit square .
the typical initial and final distributions of pvs are shown in figures 1 - 4 .
the corresponding @xmath2 functions can be found in @xcite . [
cols="^,^ " , ]
clustering of pvs is studied numerically based on point process theory and the @xmath2 functions are computed from the final distributions of pvs . clustering persists up to 10 eddy turnover time .
we also did a numerical simulation for slightly perturbed square vortex lattices , which can be regarded as a special circumstance of case ( iii ) .
it showed inactive motions of pvs compared with the other examples shown in this paper .
it may imply that the instability of square lattices requires more than 100 pvs , or the rate of growth is very small .
the author is grateful for professor t. yamagata for support through his research on fluid dynamics over these several years . | the motion of assemblies of point vortices in a periodic parallelogram can be described by the complex position @xmath0 whose time derivative is given by the sum of the complex velocities induced by other vortices and the solid rotation centered at @xmath1 .
a numerical simulation up to 100 vortices in a square periodic box is performed with various initial conditions , including single and double rows , uniform spacing , checkered pattern , and complete spatial randomness .
point process theory in spatial ecology is applied in order to quantify clustering of the distribution of vortices . in many cases , clustering of the distribution
persists after a long time if the initial condition is clustered . in the case of positive and negative vortices with the same absolute value of strength ,
the @xmath2 function becomes positive for both types of vortices .
scattering or recoupling of pairs of vortices by a third vortex is remarkable . *
* point process analysis of vortices in a periodic box + makoto umeki department of physics , graduate school of science university of tokyo , 7 - 3 - 1 hongo , bunkyo - ku , tokyo 113 - 0033 |
the motivation for investigating the low energy @xmath4 interaction is closely connected with understanding of the nature of scalar resonances @xmath6(980 ) and @xmath7(980 ) . besides the interpretation as @xmath8 mesons @xcite
, these particles were also proposed to be @xmath9 tetraquark states @xcite , hybrid @xmath8/meson - meson systems @xcite or even quark - less gluonic hadrons @xcite . since both @xmath6(980 ) and @xmath7(980 )
masses are very close to the sum of the @xmath10 and @xmath11 masses , they are considered also as @xmath2 bound states @xcite .
the strength of the @xmath2 interaction is a crucial quantity regarding the formation of such molecules .
+ the @xmath4 interaction was studied experimentally inter alia in the @xmath3 reaction with cosy11 and anke detectors operating at the cosy synchrotron in jlich in germany .
the experimental data collected systematically below @xcite and above @xcite the @xmath12 meson threshold revealed a significant enhancement in the shape of the excitation function near the kinematical threshold . on the other hand , despite the search done by the cosy11 experiment @xcite and analysis based on big data samples collected by anke and wasa at cosy experiments , there is no clear evidence of the @xmath4 pairs production through the @xmath6(980 ) or @xmath7(980 ) resonances . the enhancement of the excitation function near the threshold may be due to the final state interaction ( fsi ) in the @xmath13 system . indeed , the differential spectra obtained by the cosy-11 @xcite and anke @xcite groups indicate a strong interaction in the @xmath14 an @xmath15 subsystems . the phenomenological model proposed by the anke collaboration based on the factorization of the final state interaction into interactions in the @xmath16 and @xmath14 subsystems allowed to describe the experimental @xmath14 an @xmath15 invariant mass distributions assuming an effective @xmath14 scattering length @xmath17 fm @xcite .
however , the data very close to the kinematical threshold remain underestimated , which indicates that in the low energy region the influence of the @xmath18 final state interaction may be significant @xcite . motivated by this observation
the cosy11 collaboration has estimated the scattering length of the @xmath18 interaction based for the first time on the low energy @xmath19 goldhaber plot distributions measured at excess energies of q = 10 mev and 28 mev @xcite .
+ in this article we present preliminary results of the @xmath4fsi studies combining the goldhaber plot distributions established by the cosy11 group with the experimental excitation function near threshold .
the final state interaction model used in the presented analysis is was based on the factorization ansatz mentioned before , with an additional term describing the interaction of the @xmath4 pair ( the @xmath20 interaction was neglected since it has to be found weak @xcite ) .
we have assumed that the overall enhancement factor originating from final state interaction can be factorized into enhancements in the proton
proton , the two @xmath14 and the @xmath4 subsystems : @xmath21 where @xmath22 stands for the relative momentum of particles in the corresponding subsystem .
the proton proton scattering amplitude was taken into account using the following parametrization : @xmath23 where @xmath24 stands for the square root of the coulomb penetration factor @xcite .
the parameter @xmath25 denotes the phase shift calculated according to the modified cini fubini
stanghellini formula with the wong
noyes coulomb correction @xcite .
factors describing the enhancement originating from the @xmath14 and @xmath4fsi were instead parametrized using the scattering length approximation : @xmath26 where @xmath27 is the scattering length of the @xmath18 interaction treated as a free parameter in the analysis . since the @xmath28 scattering length estimated by the anke group
should be rather treated as an effective parameter @xcite , in the analysis we have used more realistic @xmath29 value estimated independently as a mean of all values summarized in ref .
@xcite : @xmath30 ) fm .
+ it has to be stressed , that within this simple model we neglect the charge exchange interaction allowing for the @xmath31 transitions , and generating a significant cusp effect in the @xmath18 invariant mass spectrum near the @xmath32 threshold @xcite .
however , the anke data can be described well without introducing the cusp effect @xcite , thus we neglect it in this analysis .
we also can not distinguish between the isospin i = 0 and i = 1 states of the @xmath4 system .
however , as pointed out in @xcite , the production with i = 0 is dominant in the @xmath33 reaction independent of the exact values of the scattering lengths .
- @xmath34 distribution as a function of @xmath35 and @xmath36(left ) , as well as its projections on each axis ( center and right ) .
@xmath34 denotes the absolute minimum with respect to parameters @xmath37 , @xmath35 , and @xmath38 . in the figure on the left
the area of the squares is proportional to the @xmath39 - @xmath34 value.,title="fig:",scaledwidth=30.0% ] - @xmath34 distribution as a function of @xmath35 and @xmath36(left ) , as well as its projections on each axis ( center and right ) .
@xmath34 denotes the absolute minimum with respect to parameters @xmath37 , @xmath35 , and @xmath38 . in the figure on the left
the area of the squares is proportional to the @xmath39 - @xmath34 value.,title="fig:",scaledwidth=30.0% ] - @xmath34 distribution as a function of @xmath35 and @xmath36(left ) , as well as its projections on each axis ( center and right ) .
@xmath34 denotes the absolute minimum with respect to parameters @xmath37 , @xmath35 , and @xmath38 . in the figure on the left
the area of the squares is proportional to the @xmath39 - @xmath34 value.,title="fig:",scaledwidth=30.0% ]
in order to estimate the strength of the @xmath4 interaction the experimental goldhaber plots , determined at excess energies of q = 10 mev and q = 28 mev @xcite , were compared together with the total cross sections to the results of the monte carlo simulations treating the @xmath4 scattering length @xmath40 as an unknown parameter .
we have constructed the following @xmath39 statistics : @xmath41 , \label{eqchi2_mh}\ ] ] where the first term was defined following the neyman s @xmath39 statistics , and accounts for the excitation function near threshold for the @xmath19 reaction .
@xmath42 denotes the @xmath43th experimental total cross section measured with uncertainty @xmath44 and @xmath45 stands for the calculated total cross section normalized with a factor @xmath37 treated as an additional parameter of the fit .
@xmath45 was calculated for each excess energy @xmath46 as a phase space integral over five independent invariant masses @xcite .
the second term of eq .
[ eqchi2_mh ] corresponds to the poisson likelihood @xmath47 @xcite describing the fit to the goldhaber plots determined at excess energies @xmath48 mev ( @xmath49 = 1 ) and @xmath50 mev ( @xmath49 = 2 ) using cosy11 data @xcite .
@xmath51 denotes the number of events in the @xmath52 bin of the @xmath53 experimental goldhaber plot , and @xmath54 stands for the content of the same bin in the simulated distributions .
the simulations were normalized with a factor defined for the @xmath53 excess energy as : @xmath55 . here
@xmath56 stands for the total luminosity @xcite and @xmath57 denotes the the total number of simulated @xmath3 events .
the @xmath47 distribution obtained after subtraction of its minimum value is presented in fig .
[ fig:2 ] as a function of the real and imaginary part of the @xmath4 scattering length .
the best fit to the experimental data corresponds to : @xmath58 with a @xmath47 per degree of freedom of : @xmath59 .
the statistical uncertainties were determined at the 70% confidence level ( c.l . ) taking into account that in the case of the three fit parameters @xcite .
systematic errors due to the uncertainties in the assumed @xmath28 scattering length were instead estimated as a maximal difference between the obtained result and the @xmath60 scattering length determined using different @xmath61 values quoted in ref . @xcite . in case of the @xmath62
the differences were negligible .
+ the final state interaction enhancement factor @xmath63 in the scattering length approximation is symmetrical with respect to the sign of @xmath64 , therefore he have determined only its absolute value .
the result of the analysis with inclusion of the interaction in the @xmath4 system described in this article is shown as the solid curve in fig .
[ fig:3 ] .
one can see that it describes quite well the experimental data over the whole energy range .
triangle and circles represent the disto and anke measurements , respectively @xcite .
the squares are results of the cosy11 @xcite measurements .
the dashed dotted , dotted and solid curves represent the energy dependence obtained assuming that there is no interaction between particles , assuming the @xmath16 and @xmath14 fsi and taking into account @xmath16 , @xmath65 and @xmath4 interaction , respectively .
the dashed and dashed dotted curves are normalized to the disto data point at q = 114 mev .
, scaledwidth=40.0% ]
a combined analysis of both total and differential cross section distributions for the @xmath66 reaction in the framework of a simple factorization ansatz allowed to estimate the @xmath4 scattering length by a factor five more precise than the previous one @xcite . however , the determined @xmath40 value is still consistent with zero , which indicates that in the @xmath13 system the interaction between protons and the @xmath67 meson is dominant .
all studies of the @xmath66 reaction suggest also that the resonant @xmath18 pair production near threshold proceeds rather through the @xmath68 resonance than through scalar @xmath69/@xmath70 mesons @xcite .
+ therefore , precise determination of the kaon
antikaon scattering length requires less complicated final states like @xmath71 , where only kaons interact strongly .
this final state can be studied for example via the @xmath72 reactions with the kloe2 detector operating at the da@xmath73ne @xmath12factory @xcite .
analysis of the invariant mass distributions obtained in this reaction would allow detailed studies of the @xmath4fsi , including the contribution from the production through scalar resonances .
thus , it would be a continuation of the @xmath69 and @xmath70 studies done so far by the kloe collaboration @xcite .
this work was supported by the polish national science center through the grants no .
0469/b / h03/2009/37 , 0309/b
/ h03/2011/40 , dec-2011/03/n / st2/02641 , 2011/01/d / st2/00748 , 2011/03/n / st2/02652 , 2011/03/n / st2/02641 and by the foundation for polish science through the mpd programme and the project homing plus bis/2011 - 4/3 . 9 d. morgan , m. r. pennington , phys . rev .
d * 48 * , 1185 ( 1993 ) .
r. l. jaffe , phys .
d * 15 * , 267 ( 1977 ) .
e. van beveren , _ et al .
_ , z. phys .
c * 30 * , 615 ( 1986 ) .
r. l. jaffe , k. johnson , phys .
b60 * , 201 ( 1976 ) .
d. lohse _ et al .
. phys . * a516 * , 513 ( 1990 ) . j. d. weinstein , n. isgur , phys .
d * 41 * , 2236 ( 1990 ) .
m. wolke , phd thesis , ikp jl-3532 ( 1997 ) . c. quentmeier _ et al .
_ , phys . lett . *
b515 * , 276 ( 2001 ) .
p. winter
_ et al . _ , phys . lett . *
b635 * , 23 ( 2006 ) .
y. maeda _ et al .
_ , phys .
c * 77 * , 015204 ( 2008 ) .
q. j. ye _ et al .
_ , phys .
c * 85 * , 035211 ( 2012 ) .
f. balestra _ et al .
_ , phys . lett . *
b468 * , 7 ( 1999 ) .
p. moskal _ et al .
_ , j. phys .
g * 29 * , 2235 ( 2003 ) .
m. silarski _ et al .
_ , phys .
c * 80 * , 045202 ( 2009 ) . c. wilkin , acta phys .
* 2 * , 89 ( 2009 ) .
p. moskal _ et al .
_ , phys .
lett . * b482 * , 356 ( 2000 ) . h. p.
noyes , h. m. lipinski , phys .
c * 4 * , 995 ( 1971 ) .
noyes , ann .
nucl . part .
* 22 * , 465 ( 1972 ) . j. p. naisse , nucl . phys . a * 278 * , 506 ( 1977 ) .
a. d. martin , nucl . phys . *
b179 * , 33 ( 1981 ) .
y. yan , arxiv:0905.4818 [ nucl - th ] .
a. dzyuba _ et al .
_ , phys . lett . *
b668 * , 315 ( 2008 ) .
nyborg _ et al .
_ , phys .
rev . * 140 * , 914 ( 1965 ) .
s. baker and r. d. cousins , nucl . instrum .
meth . * 221 * , 437 ( 1984 ) .
f. james , comput .
commun . * 20 * , 29 ( 1980 ) .
g. amelino - camelia _ et al .
_ , eur .
j. c * 68 * , 619 ( 2010 ) .
f. ambrosino _ et al .
_ , phys .
b * 679 * , 10 ( 2009 ) .
f. ambrosino _ et al .
. lett .
b * 681 * , 5 ( 2009 ) .
f. ambrosino _ et al .
_ , eur .
j. c * 49 * , 473 ( 2007 ) .
f. ambrosino _ et al .
_ , phys .
b * 634 * , 148 ( 2006 ) . | the strength of the kaon
antikaon interaction is a crucial quantity for many physics topics .
it is for example , an important parameter in the discussion on the nature of the scalar resonances @xmath0 and @xmath1 , in particular for their interpretation as a @xmath2 molecules .
so far , one of the few possibilities to study this interaction is the kaon pair production in multi particle exit channels like @xmath3 . in this article
we present the latest results of the @xmath4 interaction preformed based on near threshold data gathered at the cooler synchrotron cosy .
we discuss also shortly perspectives for a new measurement of the kaon
antikaon scattering length in the @xmath5 collisions . |
the idea of exploiting cosmic antiprotons measurements to probe unconventional particle physics and astrophysics scenarios has a long history [ 1 - 10 ] and moved the cosmologists for several decades . shortly after the discovery of the cp violation in the weak interactions in 64
, sakharov formulated his famous hypotheses that were assumed to be a reasonable starting point to explain the apparent contradiction between the fundamental laws of the nature and the observations .
several balloon borne experiments were dedicated to the search for antiparticles and antinuclei , and in the 70 s the teams of r. golden in usa @xcite and of e. bogomolov in russia @xcite identified the first antiprotons in cosmic rays ( the positrons were discovered more than 40 years before by anderson ) .
then , the antiproton spectrum was intensively studied for searching for signals exceeding the background of the antiprotons produced in the interactions of cr s with the interstellar matter . in figure [ beaf90 ] ( on the left ) are shown the data on the antiproton / proton ratio before 1990 . at that time
the standard production models ( black lines ) @xcite could not account for all the antiprotons measured by the golden et al . and
buffington et al .
@xcite experiments , and this triggered the formulation of many exotic models ranging from antiprotons coming from antigalaxies @xcite to annihilation of supersimmetric dark matter ( gray curve ) @xcite . in the same years the results of the positron ratio measurements were somewhat similar , with the experiments giving an higher flux of positron at energies greater than 10 gev
( see figure [ beaf90 ] on the right ) , explained only with some exotic production , like again the annihilation of wimps giving a contribution as shown by the gray curve in the figure [ beaf90 ] ( from @xcite ) .
in the last decade there has been an increase of experiments performed using novel techniques developed for accelerator physics .
this has permitted to improve the statistical and systematic significance of the experimental measurements .
the wizard collaboration did several balloon flights , mass89@xcite , mass91@xcite , ts93@xcite , caprice94@xcite and caprice98@xcite , exploring an energy interval from some hundreds mev to 50 gev of the antiproton and positron spectra .
the core parts of the instrument were a magnetic spectrometer composed by a superconducting magnet and a tracking system for charge sign and momentum measurements , an imaging calorimeter ( streamer tube or silicon tungsten ) and a @xmath1 selector ( cerenkov , trd or rich ) . in figure [ antip ]
there are reported the data for the antiproton / proton ratio collected up to now , together with the data of heat - pbar @xcite , heat@xcite and imax92@xcite experiments .
the solid and dashed lines are , respectively , the theoretical interstellar and solar modulated ratios for a pure secondary origin of the antiprotons .
the others are more exotic models . in figure [ antipf ]
there are the experimental results , including the ams data@xcite , for the antiproton flux along with different theoretical calculations which account for a pure secondary component@xcite and together with the distortion on the antiproton flux ( dashed line ) due to a possible contribution from neutralino annihilation ( dotted line , from@xcite ) . in figure [ pos ]
there are the experimental data for the @xmath2 fraction together with the distortion ( solid line ) due to a possible contribution from neutralino annihilation ( dotted line , from@xcite ) . the possibility to detect this kind of distortions and then the possibility to discover
the nature of the dark matter is not so exotic as it might appear . over the last years
our knowledge of the inventory of matter and energy in the universe has improved dramatically .
astrophysical measurements from several experiments are now converging and a standard cosmological model is emerging .
for this model the total matter density is about 40%@xmath310% of the critical density of the universe , with a contribution of the baryonic dark matter less than 5% and a contribution from neutrinos that can not be greater than 5% .
the remaining matter should be composed of yet - undiscovered weakly interacting massive particles ( wimp ) , and a good candidate for wimp s is the lightest supersymmetric particle ( lsp ) in r - parity conserving supersymmetric models . the best candidate as lsps are the neutralinos .
they are majorana fermions and annihilate with each other in the galactic halo producing leptons , quarks , gluons , gauge bosons and higgs bosons
. these decay and/or form jets that will give rise to antiprotons ( and antineutrons which decay shortly to antiprotons ) , positrons and gammas . a description of the signature of the annihilation in gamma spectra can be found in @xcite . from the figures [ antipf ] and [ pos ]
it is clear that new experiments with better statistic are needed to discriminate among the models .
an other goal of these experimens is the search of antinuclei .
detection of heavy ( z@xmath42 ) antimatter of primary origin in cosmic rays would be a discovery of fundamental significance . that would provide direct evidence of the existence of antimatter in the universe .
the current theory suggests that the remaining matter is the remnant of the almost complete annihilation of matter and antimatter at some early epoch , which stopped only when there was no more antimatter to annihilate .
starting from a matter / antimatter symmetric universe , the required conditions for a following asymmetric evolution are the cp violation , the baryon number non - conservation and a non equilibrium environment . on the basis of gamma - ray observations , the coexistence of condensed matter and antimatter on scales smaller than that of clusters of galaxies has been virtually ruled out .
however , no observations presently exclude the possibility that the domain size for establishing the sign of cp violation is as large as a cluster or super - cluster of galaxies .
for example , there could be equality in the number of super - clusters and anti - super - clusters .
similarly , there is nothing that excludes the possibility that a small fraction of the cosmic rays observed at earth reaching our galaxy from nearby super - clusters .
up to now no antihelium or heavier antinucleus have been found ; in figure [ antinu ] there are the present experimental limits and the sensitivities foreseen for the next experiments that will be discussed in the next paragraphs .
three major experiments are being prepared for the future : the polar balloon bess instrument , the satellite pamela experiment and ams on the international space station .
the pamela telescope , shown in figure [ pamd ] , is a satellite - borne apparatus built by the wizard - pamela collaboration @xcite .
the list of the people and the institution involved in the collaboration together with the on - line status of the project is available at _
the pamela scientific primary goal is the search for heavy nuclei and non baryonic particles outside the standard model , for the understanding of the formation and evolution of our galaxy and the universe and for the exploring of the cycles of matter and energy in the universe .
additional objectives of pamela are the studies of galactic cosmic rays in the heliosphere , solar flares , distribution and acceleration of solar cosmic rays ( scr s ) in the internal heliosphere , magnetosphere and magnetic field of the earth , stationary and disturbed fluxes of high energy particles in the earth s magnetosphere and anomalous components of cosmic rays . 0.2 cm the pamela observations will extend the results of balloon - borne experiments over an unexplored range of energies with unprecedented statistics and will complement the information gathered from the great space observatories . more precisely , during its three years of planned operation , pamela will measure with very high statistics : @xmath5 positron flux from 50 mev to 270 gev ( present limits 0.7 - 30 gev ) @xmath5 antiproton flux from 80 mev to 190 gev ( present limits 0.4 - 50 gev ) @xmath5 limit on antinuclei @xmath6 ( present limit about @xmath7 ) @xmath5 electron flux from 50 mev to 3tev @xmath5 proton flux from 80 mev to 700 gev @xmath5 light nuclei flux ( up to oxygen ) from 100 mev / n to 200 gev / n @xmath5 electron and proton components up to 10 tev in addition it will assure a continuous monitoring of the cosmic rays solar modulation 0.2 cm as an example , the expected data from the experiment pamela in the annihilation scenario for three years of operation are shown by black squares in figure [ pamela2 ] for both the positron and antiproton fluxes . the pamela telescope , based on the experience gained in the wizard balloon flights and in the wizard@xcite on astromag and mass - sat @xcite proposals , is composed by : 0.2 cm \1 . a magnetic spectrometer to determine the sign of the electric charge and to measure the momentum of the particles .
the magnetic system is composed by five permanent magnet of nd - fe - b , each 8 cm high , that provide a field inside the tracking volume of about 0.4 t. there are six planes of silicon micro - strip detectors for tracks reconstruction with a spatial resolution of 3 @xmath8 m in the bending view giving a maximum detectable rigidity ( mdr ) of 740 gv / c .
@xmath5 a transition radiation detector ( trd ) for distinguishing at the level of @xmath9 between electromagnetic and hadronic showers from 1 gev to about 1 tev ; the trd is based on small diameter straw tubes filled with xe - co@xmath10 mixtures and arranged in double layer planes interleaved by carbon fiber radiator
. @xmath5 a @xmath11 silicon imaging calorimeter to discriminate at the level of @xmath12 between electromagnetic and hadronic showers .
it is composed by silicon strip sensors , interleaved with tungsten plates as converters .
the energy resolution for positrons is of the order of 15% / e@xmath13 . in self triggering mode the calorimeter geometric factor
is about 470 @xmath14 , allowing the measuring with useful statistics of the electron flux up to 2 tev .
@xmath5 six scintillation counter ( each 7 mm thick ) hodoscopes for the event trigger and for tof measurements with a time resolution of @xmath15 110 ps to provide albedo discrimination and particle identification up to 3 gev / c .
@xmath5 an additional plastic scintillator ( s4 in the figure[pamela2 ] ) and a neutron detector composed by 36 @xmath16he counters in a polyetilen moderator allow , together with the imaging calorimeter , to extend the energy range for primary protons and electrons up to 10 tev .
@xmath5 a set of scintillation counters , covering part of the top edge , the lateral sides of the magnetic spectrometer and the bottom part of the calorimeter , completes the telescope , for a further labeling of contaminating events .
others characteristics of the pamela instrument are : \1 . an acceptance ( geometrical factor ) of 20.5 @xmath17 sr ; \2 . a total volume of @xmath18 cm@xmath16 ; \3 . a total mass of 470 kg ; \4 .
a power consumption of 360 w. the pamela instrument will be installed onboard the russian resurs - dk1 satellite , built by tsskb - progress .
it will be launched in 2003 from baikonur with a soyuz tm rocket and placed on an elliptic orbit at altitude 300 - 600 km and an inclination of 70.4@xmath20 for a mission at leas three years long .
averaging on the solar activity , in the first three years of the pamela flight ( 2003 - 2006 ) we expect to collect the following approximate numbers for particles , antiparticles and some nuclei : [ h ] [ cols="<,<,<,<",options="header " , ]
8 . j. silk and m. srednicki , _ phys.rev.lett . _
* 53 * ( 1984 ) 624 f.w .
stecker , s. rudaz and t.f .
walsh , _ phys.rev.lett . _
* 35 * ( 1985 ) 1 f.w .
stecker and a.j .
tylka , _ astophys . journal _ * 336 * ( 1989 ) l51 g. basini , a. morselli , m. ricci , _
la rivista del nuovo cimento _ * 12 * ( 1989 ) 4 j. silk , and a. stebbins , _ astophys . journal _ * 411 * ( 1993 ) 439 g. jungman and m. kamionkowski , _ phys.rev . _
* d51 * ( 1995 ) 3121 a. morselli _ the dark side of the universe _ , world scientific co. ( 1994 ) 267 t. mitsui et al .
lett . _ * b389 * ( 1996 ) 169 a. morselli , 8@xmath21 marcel grossmann meeting , jerusalem ( 1997 ) 1465 a. bottino , f. donato , n. fornengo , p. salati , _ phys .
* d58 * ( 1998 ) 123503 r. golden et al .
* 43 * ( 1979 ) 1196 ; see also r. golden et al .
, _ astophys .
journal lett .
_ , * 24 * ( 1984 ) 75 e. bogomolov et al .
, _ xx int .
cosmic ray conf .
_ , 2 , 72 ( 1987 ) , moscow + _ xxi int
. cosmic ray conf .
_ , 3 , 288 ( 1990 ) , adelaide r.e .
streitmatter et al
. _ xxi int .
cosmic ray conf .
_ , 3 , 277 ( 1989 ) , adelaide s.p .
ahlen et al . , _ phys .
lett . _ * 61 * ( 1988 ) 145 a. buffington et al .
, _ astophys . journal _ * 248 * ( 1981 ) 1179
r.l.golden et al . ; _ astophys . journal _ *
287 * 1987 622 d. mller and k. k. tang , _ astophys . journal _
* 312 * ( 1987 ) 183 a. buffington et al . ,
. journal _ * 199 * ( 1975 ) 669 j.k.daugherty et al . , 198 ( 1975 ) 493 j.l.fanselow et al .
, _ astophys . journal _
* 158 * ( 1969 ) 771 t.k .
gaisser and b.g .
mauger , _ phys.lett . _ * 30 * ( 1973 ) 1264 + t.k . gaisser and r.k .
schaefer , _ astrophys . journal _ * 394 * ( 1992 ) 174 + m. simon and u. heinbach , _ astophys . journal _ * 456 * ( 1996 ) 519 + t. mitsui et al .
lett . _ * b389 * ( 1996 ) 169 f.w .
stecker and a.w .
wolfendale , _ xix int .
cosmic ray conf .
_ og 6.1 - 8 , 354 ( 1985 ) la jolla a. j. tilka , _ phys .
* 63 * ( 1989 ) 840 r. golden et al .
inst . and met . _ * a306 * ( 1991 ) 366 m. bocciolini et al . , _
_ * b32 * ( 1993 ) 77 g. barbiellini et al .
inst . and met .
_ * a371 * ( 1992 ) 169 d. bergstrm et al . , _ nucl .
inst . and met . _ * a463 * ( 2001 ) 161 s. orito et al . , _ phys . rev .
lett . _ * 84 * ( 2000 ) 1078 s. w. barwick et al .
, _ astophys . journal _ * 498 * ( 1998 ) 779 d. bergstrm et al .
, _ astophys . journal _
* 534 * ( 2000 ) l177 + m. boezio et al .
, _ astophys . journal _ * 561 * ( 2001 ) 787 g. basini et al .
, _ xxvi int .
cosmic - ray conf . _
* 4 * , 77 ( 1999 ) salt lake city , + m. hof et al . , _ astophys .
journal _ * 467 * ( 1996 ) l33 m. boezio et al .
, _ astophys . journal _
* 487 * ( 1997 ) 415 j. mitchell et al .
* 76 * ( 1996 ) 3057 a. s. beach et al . , astro - ph/0111094 m.aguilar , submitted to physics reports m. boezio et al .
, _ xxvi int .
cosmic - ray conf .
_ , og.1.1.16 , ( 1999 ) salt lake city j. alcaraz et al .
* b461 * ( 1999 ) 387 m. boezio et al . , _ astophys .
* 532 * ( 2000 ) 653 j. m. clem et al . , _ astophys .
journal _ * 464 * ( 1996 ) 507 r. golden et al .
, _ astophys . journal _
* 457 * 1996 l103 r. golden et al .
, _ astophys . journal _
* 436 * ( 1994 ) 769 m. simon , a. molnar and s. roesler , _ astophys . journal _ * 499 * ( 1998 ) 250 l. bergstrm , j. edsj and p. ullio , _ astophys .
journal _ * 526 * ( 1999 ) 215 p. ullio 1999 , astro - ph/9904086 and frascati physics series vol.xxiv ( 2002 ) 475 e. baltz and j. edsj , _ phys
_ * d 59 * ( 1999 ) 023511 a.morselli , _ int .
journal of modern physics _ * a17 * 12 & 13 ( 2002 ) 1829 o.adriani et al . 1999 ,
cosmic ray conference , rome , 3 , 591 , 1995 + a. morselli , p.picozza 1999 , the pamela apparatus for the search of antimatter in cosmic rays , topics in cosmic - ray astrophysics editor : m. duvernois , pg.37 .
nova science publisher .
p.spillantini et al . , _
il nuovo cimento _ * 103b * ( 1989 ) 625 g.basini et al . , _ il nuovo cimento _ * 105b * ( 1990 ) 779 | two of the most compelling issues facing astrophysics and cosmology today are to understand the nature of the dark matter that pervades the universe and to understand the apparent absence of cosmological antimatter . for both issues , sensitive measurements of cosmic - ray antiprotons and positrons , in a wide energy range , are crucial
. many different mechanisms can contribute to antiprotons and positrons production , ranging from conventional reactions like @xmath0_anything _ up to exotic processes like neutralino annihilation .
the open problems are so fundamental
( i.e. : is the universe symmetric in matter and antimatter ? ) that experiments in this field will probably be of the greatest interest in the next years
. here we will summarize the present situation , showing the different hypothesis and models and the experimental measurements needed to lead to a more established scenario .
# 1 # 2 # 3 _ mem .
soc .
astron .
it . _ * # 1 * # 2 # 3 # 1 # 2 _ the messenger _ * # 1 * , # 2 # 1 # 2 _ astron .
nach . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
lett . _ * # 1 * , l#2 # 1 # 2 _ astron .
astrophys .
rev . _ * # 1 * , # 2 # 1 # 2 # 3 _ astron
. astrophys .
suppl .
ser . _ * # 1 * # 2 # 3 # 1 # 2 _ astron . j. _ * # 1 * , # 2 # 1 # 2 _ ann . rev
. astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 # 3 _ astophys .
journal _ * # 1 * # 2 # 3 # 1 # 2 _ astrophys .. j. lett . _ * # 1 * , l#2 # 1 # 2 _ astrophys .
j. suppl .
_ * # 1 * , # 2 # 1 # 2 _ astrophys .
space sci . _ * # 1 * , # 2 # 1 # 2 _ adv .
space res .
_ * # 1 * , # 2 # 1 # 2 _ bull .
astron .
inst .
czechosl . _ * # 1 * , # 2 # 1 # 2 _ j. quant .
spectrosc .
. transfer _ * # 1*,#2 # 1 # 2 _ mon . not .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ mem .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ phys .
lett .
rev .
_ * # 1 * , # 2 # 1 # 2 _ publ .
astron .
. japan _ * # 1 * , # 2 # 1 # 2 _ publ .
astr .
soc .
pacific _ * # 1 * , # 2 # 1 # 2 _ nature _ * # 1 * , # 2 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we must regard it rather an accident + that the earth and presumably the whole + solar system contains a preponderance + of negative electrons and positive protons . +
it is quite possible that for some of the stars + it is the other way about .
+ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
consider the group @xmath9 of invertible matrices over the finite field @xmath10 .
denote by @xmath11 its subgroup of upper triangular matrices , and by @xmath12 its subgroup of diagonal matrices .
set @xmath13 in @xmath14 $ ] .
the quotient group @xmath15 is isomorphic to the symmetric group @xmath16 .
moreover , the iwohori - hecke @xmath8-algebra @xmath17\varepsilon$ ] is isomorphic to @xmath18 as a @xmath8-vector space , and the structure constants in the multiplicative table lie in @xmath2 $ ] .
more generally , if @xmath19 is a finite reductive group over @xmath20 , @xmath5 is a borel subgroup of @xmath19 , and @xmath21 a maximal torus included in @xmath5 , then @xmath22 is a weyl group and the above results extend to the hecke algebra @xmath23 .
now , consider a _ finite reductive monoid _
@xmath4 over @xmath20 as defined by renner in @xcite .
such a monoid is a unit regular monoid and its unit group is a finite reductive group @xmath19 .
solomon introduced in @xcite the notion of a _ iwahori - hecke algebra _
@xmath6 of a finite reductive monoid @xmath4 . here
, @xmath5 is a borel subgroup of @xmath19 .
this @xmath8-algebra is defined by @xmath24\varepsilon$ ] where as before @xmath25 in @xmath26 $ ] . in this framework ,
the weyl group is replaced by an inverse monoid @xmath0 , which is called the _ renner monoid _ of @xmath4 .
its turns out that @xmath6 is isomorphic to @xmath27 as a @xmath8-vector space .
an isomorphism is given by @xmath28 .
therefore , this is natural to address the question of the existence of a normalisation @xmath29 of the basis @xmath30 such that in this new basis @xmath31 , the structure constants in the multiplicative table lie in @xmath2 $ ] as in the case of finite reductive groups .
solomon considered this question in @xcite and answered in the positive in the specific case where @xmath32 . in @xcite
, he announced that in a forthcoming paper , he was going to extend his result and its proof to every finite reductive monoid that arises as the set of fixed points of a _ reductive monoid _ over @xmath33 ( see section [ sousect ] for a definition ) by the frobenius map @xmath34 defined by @xmath35 .
but it seems that this result has never be published . in @xcite
putcha proves that for every finite reductive monoid , one can normalised the basis @xmath30 such that the structure constants become rational in @xmath7 .
howewer , the question remained open , and renner concluded in @xcite that `` the delicate part here is obtaining integral structure constants '' .
the main object of this article is to answer solomon s question in the positive for every finite reductive monoid .
we prove : let @xmath4 be a finite reductive monoid over @xmath20
. denote by @xmath0 the associated renner monoid .
there exists a normalisation of the basis @xmath30 of the iwahori - hecke algebra @xmath6 such that the structure constants in the multiplicative table lie in @xmath2 $ ] .
moreover , the coefficients of the polynomials depend on @xmath0 only .
[ thintro ] in section [ sectionprincipale ] , we provide explicit formulae ( see theorem [ th : genthe ] ) , which are related to the existence of a length function on @xmath0 .
moreover , we deduce a finite presentation of @xmath6 in the spirit of the classical presentation of @xmath23 ( see corollary [ thconclu ] in section [ sousectionrappelamt ] ) . mokler , renner and putcha consider families of monoids that are closed to reductive monoids ( see @xcite for instance .
they are called _ finite monoids of lie type _ and _ face monoids_. indeed , finite reductive monoids are special cases of finite monoids of lie type .
to each of these groups can be associated a so - called _ renner monoid _ ,
whose properties are closed to renner monoids of ( finite ) reductive monoids ( see examples [ exe : ex2 ] and [ exe : ex3 ] below ) .
this explains why these monoids are still called renner monoids in the latter references .
however , there is some differences between these monoids ( see remark [ rem : diffrenner ] for a discussion ) .
we introduce here the notion of a _ generalised renner monoid_. all renner monoids are examples of generalised renner monoids .
one motivation for this definition is to introduce a notion that plays for these various renner monoids the role of the notion of a coxeter system for weyl group .
we prove that all the properties shared by the various renner monoids hold for generalised renner monoid .
in particular , it is a factorisable monoid and its unit group @xmath19 is a coxeter group .
the crucial point regarding solomon s question is that we can associate with each such generalised renner monoid @xmath0 a _ generic hecke algebra _
@xmath1 which is a ring on the free @xmath2$]-module with basis @xmath0 .
it turns out that theorem [ thintro ] is a consequence of let @xmath4 be a finite reductive monoid over @xmath20 with renner monoid @xmath0 .
the iwahori - hecke algebra @xmath6 is isomorphic to the @xmath8-algebra @xmath36 , where @xmath37 is the specialisation of the generic hecke algebra @xmath1 at @xmath7 .
[ thintro2 ] the second main ingredient used in the proof of theorem [ thintro ] is the existence of a length function @xmath38 on every generalised renner monoid @xmath0 .
this length function is related to the canonical generating set @xmath39 , which equips every generalised renner monoid . in the case of reductive monoids
, we investigate the relation of this length function with the product of double classes .
we prove in particular that let @xmath4 be a reductive monoid with unit group @xmath19 and renner monoid @xmath0 .
fix a maximal torus @xmath21 and a borel subgroup @xmath5 that contains @xmath21 in @xmath19 .
+ ( i ) let @xmath40 lie in @xmath0 and @xmath41 lie in @xmath42 , then @xmath43 [ pr : lienlongbbintro ] + ( ii ) let @xmath40 lie in @xmath0 and @xmath44 lie in @xmath45 , then then @xmath46 this result extends results obtained in @xcite , and leads to a similar result for finite reductive monoids .
the paper is organised as follows . in section
[ sec : genheckalgr ] , we introduce the notion of a _ generalised renner monoid _ , provide examples and investigate properties of such monoids . in particular , we define the length function @xmath38 and prove that a _ generic hecke algebra _ can be associated with every generalised renner monoid . in section [ sectionprincipale ] , we first recall the notion of a reductive monoid and prove proposition [ pr : lienlongbbintro ] .
then we introduce the notion of a iwahori - hecke algebra in the context of monoid theory .
we prove some motivating general results for such algebras .
these results are probably well - known by semigroup experts , but we have not be able to find references for them . finally , we turn to finite reductive monoids and conclude with the proof of theorem [ thintro ] and [ thintro2 ] .
the notion of a coxeter group has been introduced in order to study weyl groups .
our objective in this section is to develop a similar theory for renner monoids .
we need first to recall some standard notions and introduce useful notations . if @xmath4 is a monoid , we let @xmath47 and @xmath48 its idempotent set and its unit group . we see a ( _ lower _ ) _ semi - lattice _ as a commutative idempotent semigroup where @xmath49 iff @xmath50 . in particular , @xmath51 .
a semigroup is _ unit regular _
if @xmath52 , and it is _ factorisable _ if it is unit regular and @xmath47 is a semi - lattice . in this latter case
@xmath4 is _ invertible _ , that is for every @xmath53 in @xmath4 there exists a unique @xmath54 in @xmath4 such that @xmath55 ( and therefore @xmath56 ) .
let @xmath57 be a finite simple labelled graph whose labels are positive integers greater or equal than @xmath58 .
we let denote @xmath42 the vertex set of @xmath57 .
we let @xmath59 denote the set of pairs @xmath60 such that either @xmath61 is an edge of @xmath57 labelled by @xmath62 , or @xmath61 is not an edge of @xmath57 and @xmath63 .
when @xmath60 belongs to @xmath59 , we let @xmath64 denote the word @xmath65 of length @xmath62
. the _ coxeter group _
@xmath66 associated with @xmath57 is defined by the following group presentation @xmath67 in this case , one says that the pair @xmath68 is a _
coxeter system _ , and that @xmath69 is a coxeter group .
the coxeter graph is uniquely defined by the coxeter system .
let @xmath70 be a _ coxeter system_. + ( i ) let @xmath71 belong to @xmath72 .
the _ length _
@xmath73 of @xmath71 is the minimal integer @xmath74 such that @xmath71 has a word representative of length @xmath74 on the alphabet @xmath42 .
such a word is called a _ minimal word representative _ of @xmath71 .
+ ( ii ) the subgroup @xmath75 generated by a subset @xmath76 of @xmath42 is called a _ standard parabolic subgroup _ of @xmath69 .
@xcite let @xmath70 be a _
coxeter system _ with coxeter graph @xmath57 .
+ ( i ) for every @xmath77 , the pair @xmath78 is a coxeter system .
its graph @xmath79 is the full subgraph of @xmath57 spanned by @xmath76 .
+ ( ii ) for every @xmath80 and every element @xmath81 there exists a unique element @xmath82 of minimal length in the double - class @xmath83 .
furthermore there exists @xmath84 in @xmath75 and @xmath85 in @xmath86 such that @xmath87 with @xmath88 .
[ proppopa2 ] note that @xmath89 holds , in particular , when @xmath76 or @xmath90 are empty .
the element @xmath82 is said to be _
@xmath91-reduced_. in the sequel , we let @xmath92 denote the set of _ @xmath91-reduced _ elements .
note also that the pair @xmath93 is not unique in general , but it becomes unique if we require that @xmath94 is @xmath95-reduced ( or that @xmath96 is @xmath97-reduced ) .
if @xmath0 is a factorisable monoid and @xmath44 belongs to @xmath98 we let @xmath99 and @xmath100 denote the subgroups defined by @xmath101 @xmath102 the unit group @xmath103 acts on @xmath98 by conjugacy . 1 .
@xmath0 is a factorisable monoid ; 2 .
@xmath45 is both a transversal of @xmath98 for the action of @xmath103 and a sub - semi - lattice ; 3 .
@xmath105 is a coxeter system ; 4 . for every pair @xmath106 in @xmath98
there exists @xmath71 in @xmath103 and @xmath107 in @xmath45 such that @xmath108 for @xmath109 ; 5 . for every @xmath44 in @xmath45 ,
the subgroups @xmath99 and @xmath100 are standard coxeter subgroups of @xmath103 ; 6 .
the map @xmath110 is not decreasing : @xmath111 . in this case
, we say that @xmath0 is a _ generalised renner monoid_. following the standard terminology for renner monoids , we call the section @xmath45 the _ cross section lattice _ of @xmath0 , and we define the _ type map _ of @xmath0 to be the map @xmath112 defined by @xmath113 .
assume @xmath104 is a generalised renner - coxeter system .
+ ( i ) since @xmath100 is a standard coxeter subgroup of @xmath99 , we have @xmath116 moreover , this is clear that @xmath100 is a normal subgroup of @xmath99 . as a consequence , @xmath117(ii ) below
, several results can be proved without assuming property ( ecs6 ) .
however this is a crucial tool in the proof of theorem [ th : genthe ] and proposition [ prop : inverprese ] .
+ ( iii ) if @xmath98 is finite and a lower semi - lattice , then it has to be a lattice .
this is so for _ renner monoids _ associated with _
reductive monoids_. + ( iv ) the map @xmath118 is not increasing : @xmath119(v ) we can have @xmath120 and @xmath121 for @xmath122 ( see ( * ? ? ?
[ remdepar ] let @xmath4 be a _ abstract finite monoid of lie type _
( see @xcite,@xcite or @xcite for a definition .
note that these groups are called _ regular split monoids _ in @xcite , and _
finite monoids of lie type _ in @xcite ) . the associated _ renner monoid
_ @xmath123 of @xmath4 is a generalised renner monoid .
[ exe : ex2 ] property ( ecs6 ) follows from ( * ? ? ?
3.5(i ) ) .
the other defining properties hold by ( * ? ? ?
* sec . 2 ) .
the seminal examples of an abstract finite monoid of lie type is a renner monoid of a _ finite reductive monoid _ @xcite . in section 3
we focus on these monoids .
[ exe : ex3 ] let @xmath19 be a kac - moody group over a field @xmath124 of characteristic zero whose derived group is the special kac - moody group introduced in @xcite
. denote by @xmath70 the associated coxeter system .
the coxeter group @xmath69 is infinite .
let @xmath125 be the set of _ faces _ of its associated tits cone @xmath126 ( see @xcite for details ) .
the action of @xmath69 on @xmath126 induces an action on the lattice @xmath125 .
renner monoid _
@xmath0 is the monoid @xmath127 where @xmath128 is the congruence on @xmath129 defined by @xmath130 if @xmath131 and @xmath132 fixes @xmath0 pointwise @xcite .
then @xmath0 is a generalised renner monoid .
properties ( ecs1 ) , ( ecs2 ) , ( ecs3 ) and ( ecs5 ) are proved in @xcite ( see also @xcite ) .
the cross section lattice @xmath45 can be identified with the set of infinite standard parabolic subgroups of @xmath69 that have no finite proper normal standard parabolic subgroups .
the semi - lattice structure is given by @xmath133 if @xmath134 .
if @xmath135 belongs to @xmath45 , then @xmath136 and @xmath137 .
the latter equality clearly implies ( ecs6 ) . finally , property ( ecs4 )
can be deduced from ( * ? ? ?
* theorem 2 and 4 ) .
[ rem : diffrenner ] in examples [ exe : ex1 ] , [ exe : ex2 ] and [ exe : ex3 ] we provide examples of generalised renner monoids that are all called _ renner monoid _ in the literature . from our point of view , this is not a suitable terminology since there is crucial differences between these monoids .
therefore , using the same terminology may be misleading . for instance
, for renner monoids of reductive monoids one has @xmath138 and @xmath139 .
this is not true in general for renner monoids associated with abstract finite monoids of lie type ( see @xcite for a details ) . in renner monoids of reductive monoids and of abstract finite monoids of lie type ,
all maximal chains of idempotents have the same size . this is not true for renner monoids of example [ exe : ex3 ] , as explained in @xcite .
for all this section , we fix a generalised renner - coxeter system @xmath104 .
we let @xmath69 denote the unit group of @xmath0 .
our objective is to prove that important properties shared by renner monoids of examples [ exe : ex1 ] , [ exe : ex2 ] , [ exe : ex3 ] can be deduced from their generalised renner - coxeter system structure . in particular , we extend to this context the results obtained in @xcite . by proposition [ proppopa2 ] , for every @xmath71 in @xmath72 and every @xmath140 in @xmath45 , each of the sets @xmath141 , @xmath142 , @xmath143 , @xmath144 and @xmath145 has a unique element of minimal length . in order to simplify notation ,
we set @xmath146 , @xmath147 ; @xmath148 ; @xmath149 ; @xmath150 .
[ fnrenner ] for every @xmath40 in @xmath0 , + ( i ) there exists a unique triple @xmath151 with @xmath152 , @xmath153 and @xmath154 such that @xmath155 ; + ( ii ) there exists a unique triple @xmath156 with @xmath152 , @xmath157 and @xmath158 such that @xmath159 let us prove ( i ) .
the proof of ( ii ) is similar .
let @xmath40 belong to the monoid @xmath0 . by property ( ecs1 )
, there exists @xmath44 in @xmath98 and @xmath71 in @xmath69 such that @xmath160 . by property ( ecs2 )
there exists @xmath161 in @xmath45 and @xmath162 in @xmath69 such that @xmath163 .
then @xmath164 with @xmath165 . by remark [ remdepar](i ) , we can write @xmath166 and @xmath167 with @xmath168 , @xmath85 , @xmath169 , @xmath170 and @xmath171 in @xmath172 , @xmath173 , @xmath100 , @xmath174 and @xmath100 , respectively
. then we have @xmath175 , and @xmath176 belongs to @xmath172 , still by remark [ remdepar](i )
. now assume @xmath177 with @xmath140 in @xmath45 , @xmath178 in @xmath172 and @xmath179 , respectively , and @xmath180 in @xmath173 and in @xmath181 , respectively . then @xmath182 .
this implies @xmath183 by @xcite .
as a consequence , @xmath184 and @xmath185 lies in @xmath99 .
since @xmath186 and @xmath85 both belong to @xmath173 , we must have @xmath187 .
now , it follows that @xmath188 and @xmath189 lies in @xmath100 .
this implies @xmath190 in @xmath172 .
let @xmath140 belong to @xmath45 and @xmath71 lie in @xmath191 .
+ [ lemclef ] ( i ) there exists @xmath192 in @xmath45 such that @xmath71 belongs to @xmath193 and @xmath194 .
+ ( ii ) the element @xmath71 lies in @xmath195 .
therefore , @xmath196 .
the proof is similar to ( * ? ? ?
* prop 1.21 ) .
( i ) consider the normal decomposition @xmath199 of @xmath200 . by definition
@xmath84 belongs to @xmath201 and @xmath85 belongs to @xmath202 .
the element @xmath203 is equal to @xmath204 and belongs to @xmath98 .
since @xmath85 lies in @xmath202 , this implies that @xmath205 lies in @xmath195 , and that @xmath206 . by property ( ecs4 ) ,
there exists @xmath207 in @xmath72 and @xmath208 in @xmath45 , with @xmath209 , such that @xmath210 and @xmath211 . since @xmath45 is a cross section for the action of @xmath72 , we have @xmath212 and @xmath213 . in particular
, @xmath207 belongs to @xmath214 .
since @xmath85 belongs to @xmath202 , we deduce that there exists @xmath40 in @xmath193 such that @xmath215 with @xmath216 .
then @xmath85 lies in @xmath214 , too .
now , write @xmath217 where @xmath218 lies in @xmath219 and @xmath220 belongs to @xmath221 .
one has @xmath222 , and @xmath223 lies in @xmath224 . by symmetry
, we get that @xmath220 belongs to @xmath99 .
the element @xmath225 is equal to @xmath226 and belongs to @xmath193 .
but , by hypothesis @xmath71 lies in @xmath191
. then we must have @xmath227 . since @xmath226 belongs to @xmath193
, it follows that @xmath220 and @xmath85 belong to @xmath193 too .
this implies @xmath228 and @xmath229 .
therefore , @xmath230 .
+ ( ii ) this is a direct consequence of the following fact : for @xmath231 in @xmath45 such that @xmath232 , we have @xmath233 and @xmath234 .
assume @xmath71 lies in @xmath235 , then we can write @xmath236 where @xmath84 lies in @xmath195 and @xmath85 lies in @xmath219 .
since @xmath237 , we have @xmath238 and @xmath239 .
since @xmath71 belongs to @xmath240 , this implies @xmath241 .
the proof of the second inclusion is similar .
\(i ) for every chain @xmath242 in @xmath98 there exists @xmath71 in @xmath103 and a chain @xmath243 in @xmath45 such that @xmath108 for every index @xmath244 .
+ ( ii ) if @xmath45 has an infimum @xmath44 , then @xmath245 .
+ ( iii ) for all @xmath140 in @xmath45 and @xmath71 in @xmath191 , one has @xmath246
\(i ) assume @xmath248 .
we prove the result by induction on @xmath62 . for @xmath249
this is true by property ( ecs4 ) .
assume @xmath250 .
by induction hypothesis , we can assume @xmath251 .
we can also also assume that @xmath84 belongs to @xmath252 .
by hypothesis , we have @xmath253 . we can write @xmath254 with @xmath168 in @xmath255 , @xmath186 in @xmath256 and @xmath257 in @xmath258 . then @xmath259 . if @xmath260 , then we get a contradiction by lemma [ lemclef](i ) and proposition [ fnrenner ] . then @xmath261 and @xmath262 .
it follows that @xmath263 .
write @xmath264 and @xmath265 with @xmath266 in @xmath267 and @xmath268 in @xmath269 .
we have @xmath270 .
since @xmath271 and @xmath272 , we get that @xmath273 and @xmath274 lie in @xmath255 and @xmath256 , respectively
. then @xmath275 and @xmath276 with @xmath277 . but @xmath278 for @xmath279 .
therefore , @xmath280 for every @xmath281 .
+ ( ii ) if @xmath282 does not belong to @xmath283 , then @xmath284 in @xmath45 .
+ ( iii ) this is clear that @xmath198 lies in @xmath285 . now ,
if @xmath286 verifies @xmath237 , @xmath287 , and @xmath288 , then @xmath289 .
therefore , @xmath290 .
the last equality follows form the fact that @xmath291 belongs to @xmath292 . for every @xmath71 in @xmath69 , we fix an arbitrary reduced word representative @xmath293 .
we set @xmath294 .
the monoid @xmath0 admits the monoid presentation whose generating set is @xmath295 and whose defining relations are : an analogue of a reductive algebraic monoid , whose unit group is a kac - moody group .
823 ( 2005 ) 90 pp .
the maximal chains of the extended bruhat orders on the @xmath296-orbits of an infinite renner monoid . 35 ( 2007 ) 2298 - 2323 . | we associate with every renner monoid @xmath0 a _ generic hecke algebra _
@xmath1 over @xmath2 $ ] which is a deformation of the monoid @xmath3-algebra of @xmath0 .
if @xmath4 is a finite reductive monoid with borel subgroup @xmath5 and associated renner monoid @xmath0 , then we obtain the associated iwahori - hecke algebra @xmath6 by specialising @xmath7 in @xmath1 and tensoring by @xmath8 over @xmath3 , as in the classical case of finite algebraic groups .
this answers positively to a long - standing question of l. solomon .
_ 2000 mathematics subject classification _
: 20g40 , 20c08 , 20g05 .
+ keywords : renner monoid , hecke algebra . |
while the supergravity and string theory are well - motivated candidates of the fundamental theory , there often appear light scalar fields , called moduli , having planck - scale suppressed interactions with matter .
the moduli cause serious cosmological problems , since the lifetime is typically longer than 1 sec and the energy density stored in the form of coherent oscillation is too much .
the decay of moduli usually injects significant amount of energy after the big - bang nucleosynthesis ( bbn ) begins , and it modifies subsequent cosmological scenarios .
this is called the cosmological moduli problem @xcite .
even if the moduli are heavy enough to decay before bbn , gravitinos produced by the moduli decay again may be problematic @xcite .
thermal inflation @xcite is a short period of inflation which takes place well after the primordial inflation .
it is regarded as a solution to the cosmological moduli problem , since the coherent oscillation of the moduli is sufficiently diluted by the exponential expansion during thermal inflation and subsequent entropy - production by the flaton decay .
thermal inflation is driven by a scalar field , called _ flaton _ , having a flat potential and trapped at the origin of the scalar potential due to thermal effects .
this kind of scalar field can be embedded into supersymmetric ( susy ) theories .
however , the most thermal inflation models suffer from a problematic domain wall ( dw ) formation after the phase transition in association with the end of thermal inflation @xcite .
this is because the flaton usually has @xmath0 symmetry which is spontaneously broken at the true vacuum .
once formed , dws eventually dominate the energy density of the universe resulting in a cosmological disaster .
thus the @xmath0 symmetry , which guarantees the flatness of the flaton potential , must be explicitly broken to some extent in order for the dws to disappear before they come to dominate the universe .
actually , an introduction of a small explicit breaking in the scalar potential can solve the dw problem @xcite .
such dws can be a source of gravitational waves ( gws ) . in this paper
, we study the gws from the dynamics of dws produced after thermal inflation . we show that the amount of the gws may be within the reach of the pulsar timing experiments and future space - borne gw detectors .
thus the study of gws provides a way to probe the period of thermal inflation .
it is also shown that the @xmath0 can be naturally broken by the qcd instanton effect in a simple class of thermal inflation model ; in such a model , the abundance of the gws will be well within the observable range .
let us start with the model of thermal inflation .
we introduce a gauge singlet superfield @xmath1 which takes the role of the flaton .
imposing @xmath0 symmetry , under which @xmath1 transforms as @xmath2 , we adopt the following superpotential are also allowed , but they have little effects on the flaton dynamics discussed below and hence are neglected . ]
@xmath3 here , @xmath4 and @xmath5 are additional `` quarks '' which have quantum numbers for su(3)@xmath6 ; they are thermalized when @xmath7 , and give a thermal mass to @xmath1 .
in addition , @xmath8 is a constant which is related to the gravitino mass @xmath9 as @xmath10 ( with @xmath11 being the reduced planck scale ) . in the following , in order to guarantee the flatness of the potential of @xmath1 , we take @xmath12
. we also comment here that , if @xmath13 , the superpotential has an @xmath14-symmetry under which @xmath1 has a charge @xmath15 .
although such an @xmath14-symmetry is explicitly broken by the constant term @xmath8 , we will be interested in the case that the effect of the breaking is relatively small .
then , in the vacuum where @xmath16 , quasi nambu - goldstone boson , which we call @xmath14-axion , shows up
. including the susy breaking effect , the zero temperature potential of the flaton @xmath1 takes the following form : @xmath17 if there are no sources for the @xmath18-parameter other than @xmath8 , we have @xmath19 . the susy breaking mass parameter for @xmath1
is represented by @xmath20 and is assumed to be negative .
hereafter we take @xmath18 real and positive . in addition , we take @xmath21 for simplicity .
the vacuum expectation value ( vev ) of @xmath1 is given by @xmath22 the flaton mass around the minimum is given by @xmath23 , and @xmath24 is determined as @xmath25 so that the cosmological constant vanishes at the vacuum . in the gravity - mediated susy breaking model , @xmath26 is expected to be of the order of tev . in the gauge - mediated susy breaking model @xcite , on the other hand , the scalar potential is slightly more complicated @xcite and the parameter dependence changes .
hereafter , we treat @xmath27 and @xmath28 as free parameters without specifying the mechanism of susy breaking .
suppose that @xmath1 is trapped at the origin due to the large hubble mass term during / after inflation .
in such a period , @xmath4 and @xmath5 are massless and hence they participate to thermal bath .
thus the flaton obtains a thermal mass of @xmath29 .
as the cosmic temperature decreases , the energy density of the universe is dominated by the potential energy of @xmath1 and thermal inflation begins .
thermal inflation ends at the temperature @xmath30 for @xmath31 .
then the flaton oscillates around the minimum , and finally it decays into radiation .
we parametrize the flaton decay rate as @xmath32 with a numerical constant @xmath33 , which depends on the decay mode .
if @xmath1 decays into the higgsino pair via the superpotential of @xmath34 ( where @xmath35 and @xmath36 are the up- and down - type higgses , respectively ) with @xmath37 , then @xmath38 . on the contrary ,
@xmath39 for the decay into gluons via the loop effect .
the reheating temperature after the flaton decay is estimated as - axion pair .
if the decay into the @xmath14-axion is dominant , the final reheating completes by the decay of @xmath14-axion .
this situation is effectively incorporated in the parametrization ( [ gamma ] ) by choosing the value of @xmath33 appropriately .
in addition , the flatino , the fermionic component of the flaton , has a mass of @xmath40 , and hence the flaton can not decay into the flatino pair . ] @xmath41^{1/4 } \left ( \frac{m_\phi}{1{\rm tev } } \right)^{3/2 } \left ( \frac{10^{10}{\rm gev}}{v } \right ) , \label{tr}\ ] ] where @xmath42 is the effective number of massless degrees of freedom at the temperature @xmath43 . notice that @xmath44 is required to be higher than a few mev for successful bbn .
let us estimate the modulus abundance in the presence of thermal inflation .
we denote the modulus field by @xmath45 and its mass and energy density by @xmath46 and @xmath47 .
thermal inflation efficiently dilutes the moduli through the late - time entropy production .
the dilution factor is defined as the ratio of the entropy with and without the additional entropy production at the flaton decay .
it is given by @xcite .
@xmath48 \left ( \frac{1{\rm tev}}{m } \right ) \left ( \frac{1{\rm gev}}{t_{\rm r } } \right ) \left ( \frac{v}{10^{10}{\rm gev } } \right)^2 .
\end{split}\ ] ] the modulus abundance after the entropy production is @xmath49 where @xmath50 is the initial amplitude of the modulus .
here we have assumed that the modulus begins to oscillate before the inflaton decays and also that the inflaton decays before thermal inflation starts .
otherwise , the expression becomes more complicated @xcite .
we call this `` primary '' moduli for the reason discussed below
. there is another contribution to the modulus oscillation @xcite .
since the potential minimum of the moduli during thermal inflation may be displaced from the true minimum due to the hubble mass correction , the secondary oscillation is induced after thermal inflation ends .
the amplitude is estimated as @xmath51 .
hence the moduli abundance is estimated to be @xmath52 here @xmath53 is the true minimum of the modulus while the origin is chosen so that the hubble mass correction is given by @xmath54 .
this is called the `` secondary '' moduli .
the final modulus abundance is the sum of the primary and secondary moduli .
which one dominates the moduli density depends on various parameters .
bbn , cosmic microwave background , and diffuse @xmath55- and @xmath56-ray background give stringent upper bound on the modulus abundance , depending on its mass .
complete analyses on the modulus abundance and its cosmological effects can be found in ref .
in this section we discuss the problematic dw formation after thermal inflation , and a solution to the dw problem .
after the decay , the flaton randomly falls into one of the @xmath57-degenerate minima .
since the vev of @xmath1 spontaneously breaks the @xmath58-symmetry , dws are formed after thermal inflation .
the dw tension is estimated as @xmath59 where @xmath60 and @xmath61 gives the @xmath14-axion mass .
dws obey the scaling low in which about one dw exists per hubble horizon as far as the viscosity on the dw is negligible @xcite . and @xmath62 cause friction , but this effect is found to be insignificant for the following analysis .
] then dws begin to dominate the universe at @xmath63 this is problematic unless @xmath64 , since otherwise dws dominate the universe before the present epoch @xcite .
thus there should be an explicit @xmath58 breaking term which makes dws unstable .
let us denote the @xmath0-breaking scalar potential by @xmath65 , whose possible origins will be discussed later . it generates a bias for the @xmath57 minima of the original potential and the degeneracy among those minima are lifted completely .
dws are not absolutely stable under the existence of the bias , and eventually collapse when the bias energy density becomes comparable to the dw energy density @xcite .
this happens when @xmath66 therefore , in order for the dws to collapse before they come to dominate ( @xmath67 ) , we need the following condition , @xmath68 the dw problem is solved if this condition is satisfied .
now we describe possible origins of the bias term .
let us introduce the following superpotential @xmath69 in addition to ( [ w ] ) , where @xmath57 and @xmath70 are relatively prime numbers and @xmath71 is a small coefficient .
in order for this term not to change the flaton potential significantly , we require @xmath72 then the following additional scalar potentials are generated , @xmath73 the second term dominates for @xmath74 and this yields the bias for the @xmath57 minima of the original potential as @xmath75 the original @xmath0 symmetry is explicitly violated by the bias , and hence the degeneracy of the @xmath57 minima of the scalar potential is broken .
the condition ( [ veps_sigma ] ) is rewritten as @xmath76 if this condition is satisfied , dws collapse before they come to dominate the universe and there is no dw problem .
rather simple scenario to break the @xmath0 symmetry is to use the qcd instanton effects @xcite .
this is economical in the sense that the term @xmath77 in ( [ w ] ) , which is needed for giving rise to a thermal mass for the flaton field , also works as a source of the bias for eliminating the dws .
have a charge of _ hidden _ qcd and its scale is around gev - tev . ] the @xmath0 symmetry in ( [ w ] ) has an anomaly for su(3)@xmath6 .
consequently , the qcd instanton effect works as a bias , which lifts the classical degeneracy of the @xmath57 vacua
. the bias due to this effect is estimated to be @xcite @xmath78 for temperature below @xmath79mev , where @xmath80 and @xmath81 are the pion decay constant and mass , and @xmath82 is the current mass of the up ( down ) quark . at higher temperature ,
this effect is suppressed by high powers of @xmath43 @xcite , and hence the bias due to the qcd instanton effect is only turned on at @xmath83 @xcite .
we need two conditions for this mechanism to work successfully .
one is eq .
( [ veps_sigma ] ) for dws collapse due to the bias .
the other is @xmath84 , since otherwise the qcd instanton effect is highly suppressed and it is not suitable for the bias .
these two conditions are written as @xmath85 for @xmath86 .
therefore , for example , @xmath87gev and @xmath88gev are good choices for @xmath89 . in this choice
we have @xmath90 .
as will be seen , this predicts observable gws .
it should be noticed that @xmath91 can not take an arbitrary value in order to solve the cosmological moduli problem .
as seen in eqs .
( [ mod - prim ] ) and ( [ mod - sec ] ) , the moduli abundance crucially depends on @xmath27 , which is determined by the cutoff scale @xmath91 . comparing with
the bbn bound @xmath92gev for the moduli of @xmath93tev @xcite , @xmath94gev for @xmath89 may be allowed for moderate parameter choices .
now we are at the position to discuss the main subject of this paper , which is the gw production from dws .
we will show that the gws from the dws after thermal inflation may be observable by future experiments .
the dw network has complicated structure .
large amount of energy is localized around dws , which becomes the source gws . as a result
, considerable amount of gws can be emitted by the relativistic motion of dws .
frequencies of the gws correspond to typical scales of the dw motion , which ranges from the hubble scale to the size of dw width .
detailed calculations of the gw spectrum from dws using three dimensional lattice simulation were performed in refs .
according to recent studies @xcite , gws have rather broad spectrum which extends from the horizon size at the dw collapse for lower frequency side to the dw width for the higher frequency side . the gw spectrum , in terms of @xmath95 ( where @xmath96 is the gw energy density with frequency @xmath97 measured at present and @xmath98 is the present critical energy density ) is almost flat between these frequencies
. the energy density of gravitational waves at the collapse of dws can be estimated as @xcite @xmath99 where @xmath100 is the newton constant . here
we have substituted @xmath101 .
then , @xmath102 is given by @xmath103 \left [ \frac{g_{*s0}}{g_{*s}(t_{\rm dec } ) } \right]^{4/3 } \left ( \frac{h_{\rm dom}}{h_{\rm dec } } \right)^2 & { \rm ~~for~~ } h_{\rm r } > h_{\rm dec } , \\
\displaystyle \omega_{\rm gw}(f ) \simeq \omega_{\rm r } \left [ \frac{g_*(t_{\rm dec})}{g_{*0 } } \right ] \left [ \frac{g_{*s0}}{g_{*s}(t_{\rm dec } ) } \right]^{4/3 } \left ( \frac{h_{\rm dom}}{h_{\rm dec } } \right)^2 \left ( \frac{h_{\rm r}}{h_{\rm dec } } \right)^{2/3 } & { \rm ~~for~~ } h_{\rm r } < h_{\rm dec},\end{aligned}\ ] ] for @xmath104 defined below , where @xmath105 is the present radiation energy with three massless neutrino species density
. numerically we obtain @xmath106^{1/3 } \left ( \frac{h_{\rm dom}}{h_{\rm dec } } \right)^2 & { \rm ~~for~~ } h_{\rm r } > h_{\rm dec } , \\ \displaystyle \omega_{\rm gw}(f ) \simeq
2\times 10^{-5 } \left [ \frac{228.75}{g_*(t_{\rm dec } ) } \right]^{1/3 } \left ( \frac{h_{\rm dom}}{h_{\rm dec } } \right)^2 \left ( \frac{h_{\rm r}}{h_{\rm dec } } \right)^{2/3 } & { \rm ~~for~~ } h_{\rm r } < h_{\rm dec},\end{aligned}\ ] ] for @xmath104 . the typical frequency at the lower side , corresponding to the horizon size at the dw collapse , called the `` edge '' frequency ( @xmath107 ) in ref .
@xcite , is given by @xmath108 which is estimated as @xmath109^{1/6 } \left ( \frac{t_{\rm dec}}{1{\rm tev } } \right ) & { \rm ~~for~~ } h_{\rm r } > h_{\rm
dec } , \\
\displaystyle f_{\rm edge } \simeq 3\times 10^{-5}{\rm hz } \left [ \frac{g_*(t_{\rm r})}{228.75 } \right]^{1/6 } \left ( \frac{t_{\rm r}}{1{\rm tev } } \right ) \left ( \frac{h_{\rm dec}}{h_{\rm r } } \right)^{1/3 } & { \rm ~~for~~ } h_{\rm r } < h_{\rm dec},\end{aligned}\ ] ] where @xmath110 is the temperature at which the dws collapse . on the other hand , the `` peak '' frequency ( @xmath111 ) @xcite , corresponding to the dw width at the collapse redshifted to the present time ,
is given by @xmath112 where @xmath113 is the dw width .
this is estimated as @xmath114^{1/3 } \left ( \frac{m_a}{1{\rm tev } } \right ) \left ( \frac{1{\rm tev}}{t_{\rm dec } } \right ) & { \rm ~~for~~ } h_{\rm r } > h_{\rm dec } , \\ \displaystyle f_{\rm peak } \simeq 1\times 10^{10}{\rm hz } \left [ \frac{228.75}{g_*(t_{\rm r } ) } \right]^{1/3 } \left ( \frac{m_a}{1{\rm tev } } \right ) \left ( \frac{1{\rm tev}}{t_{\rm r } } \right ) \left ( \frac{h_{\rm r}}{h_{\rm dec } } \right)^{2/3 } & { \rm ~~for~~ } h_{\rm r } < h_{\rm dec}.\end{aligned}\ ] ] we approximate that the gw spectrum is flat between @xmath115 and @xmath116 , although the simulation shows a bit preference for the blue spectrum @xcite . in that sense ,
our estimate of the gw is conservative
. it may be within the observable range of gw detectors such as pulsar timing array or space laser interferometers @xcite .
[ fig : ogw ] shows the edge frequency @xmath107 ( left ) and @xmath102 ( right ) as a function of @xmath117 , which parameterizes the epoch of the collapse of dws .
here we have taken @xmath118tev , @xmath119gev , @xmath120 , @xmath121 and @xmath122 .
it is clear that the gw energy density becomes larger for dws that collapse closer to the epoch of dw domination .
[ fig : contour ] shows contours of @xmath102 , @xmath123 and @xmath44 for @xmath89 ( top ) and @xmath124 ( bottom ) . in the gray region denoted by `` no dw '' ,
the bias is so large , i.e. @xmath125 , that no dw formation is expected .
the current limits from pulsar timing experiments @xcite and the ligo experiment @xcite do not give stringent constraints on the model .
in the same figure , we also show the prospects of the discovery reaches of up - comming projects .
projects such as ska @xcite will probe some parameter regions .
ground based gw detectors such as the advanced ligo @xcite and lcgt@xcite are also sensitive to the higher frequency gws of @xmath126 hz . here a correlation analysis of 1 year is assumed .
space based gw detector such as lisa @xcite will also be sensitive to the gws from dw collapse for wider parameter regions .
sensitivities are found in ref .
decigo @xcite may cover the whole parameter region with @xmath127 .
the scenario with the qcd instanton as a bias term corresponds to the lower - left edge of fig .
[ fig : contour ] and hence future pulsar timing and/or space - borne gw experiments will be able to detect gws .
( left ) and @xmath102 ( right ) as a function of @xmath117 .
we have taken @xmath118tev , @xmath119gev , @xmath120 , @xmath121 and @xmath122 .
, title="fig : " ] ( left ) and @xmath102 ( right ) as a function of @xmath117 .
we have taken @xmath118tev , @xmath119gev , @xmath120 , @xmath121 and @xmath122 .
, title="fig : " ] ( solid line ) , @xmath107 ( dotted line ) and @xmath44 ( dashed line ) for @xmath89 ( top ) and @xmath124 ( bottom ) . the regions explored by lisa , ska and advanced ligo are also shown .
decigo may cover the whole parameter region with @xmath127 .
here we have taken @xmath118tev , @xmath119gev and @xmath120 .
, title="fig : " ] ( solid line ) , @xmath107 ( dotted line ) and @xmath44 ( dashed line ) for @xmath89 ( top ) and @xmath124 ( bottom ) . the regions explored by lisa , ska and advanced ligo are also shown .
decigo may cover the whole parameter region with @xmath127 .
here we have taken @xmath118tev , @xmath119gev and @xmath120 .
, title="fig : " ]
some comments are in order .
it was proposed that the first order phase transition triggers the end of thermal inflation , and correspondingly bubbles of true vacuum are created @xcite .
bubbles expand into the surrounding false vacuum regions and collide with each other , generating gws with observable level @xcite .
this predicts higher frequency gws than studied in this paper and hence will provide further information on the thermal inflation model .
we also comment on possible origin of the baryon asymmetry in the model .
since thermal inflation dilutes away the pre - existing baryon asymmetry , as well as the moduli , we need some mechanisms to create baryon asymmetry after thermal inflation . in this regard , a variant type of the affleck - dine mechanism @xcite works after thermal inflation @xcite .
one might notice that the heavy quraks @xmath4 and @xmath5 may be stable and cosmologically relevant @xcite . since they are once in thermal equilibrium during thermal inflation , its abundance just after the flaton decay is given by @xmath128 even if it is less than the dm abundance , bbn constraints on the strongly - interacting relic particles are much more stringent @xcite .- hadron formation @xcite may help the situation if the flaton decays after the qcd phase transition . ] this constraint can easily be evaded if @xmath4 and @xmath5 are in complete multiplets of the gut group , which is natural from the viewpoint of gauge coupling unification .
if @xmath4 and @xmath5 are embedded into fundamental and anti - fundamental representation of su(5)@xmath129 , for example , they should be acompanied by non - colored components ( denoted by @xmath130 and @xmath131 ) .
the renormalization group effect makes the colored components heavier than non - colored ones at the low energy .
then , @xmath4 and @xmath5 decay into @xmath130 and @xmath131 via dimension five operators suppressed by the gut scale .
the decay rate is estimated to be @xmath132 , where @xmath133 is the gut scale .
the decay rate is large enough so that colored components decay before bbn .
similarly , @xmath130 splits into the electrically charged and neutral components , where the former is heavier and decays into the latter .
notice that the neutral component may be a candidate of superheavy dm .
@xcite . ]
let us consider how robust the dw formation after thermal inflation is . in the simplest model , the flatness of the flaton potential is ensured by a @xmath0 symmetry as in ( [ w ] )
. in this case dws necessarily appear .
on the other hand , one can impose a global u(1 ) symmetry to make the flaton potential flat .
the flaton potential is stabilized by introducing another singlet scalar with some u(1 ) charge or by taking account of the radiative correction to the flaton soft mass @xcite .
then , the spontaneous breaking of the u(1 ) symmetry due to the vev of the flaton results in the formation of cosmic strings instead of dws .
even so , the global u(1 ) symmetry may be anomalous and may not be a good symmetry at the quantum level .
if only a discrete subgroup @xmath134 of the u(1 ) symmetry remains at the quantum level , dws still exist @xcite . in particular , if @xmath135 , the dw problem still exists . on the contrary , if @xmath136 , dws are bounded by strings and they disappear just after the formation @xcite .
it is also possible that the u(1 ) , which guarantees the flatness of the potential , is a gauge symmetry @xcite .
the flaton takes the role of the higgs field that gives a mass for the u(1 ) gauge boson .
then , the u(1 ) should be an exact symmetry and dw does not exist . instead
, cosmic strings are formed which are less harmful than dws .
cosmic strings formed after thermal inflation has a width of the tev scale , while the tension can be much larger .
this kind of thick strings have characteristic cosmological implications as studied in refs .
@xcite .
finally , we comment on another possible source of gws , which is the dw related to the breaking of peccei - quinn symmetry @xcite . in order to avoid the dw problem , it is often the case that axion models with @xmath136 is considered .
in such a case , dws are surrounded by cosmic strings as we have mentioned ; then , dws disappear just after the production .
this happens when @xmath137 , so we expect that the gws with the frequency of @xmath138 are emitted from the dynamics of axionic dws .
we have estimated the amount of such gws , and found that @xmath102 is a few order of magnitude smaller than the sensitivity of the future pulsar timing experiments . to summarize , a class of thermal inflation models is inevitably associated with dw formation , which reintroduces a cosmological disaster .
thus we need a mechanism to make dws unstable .
if the dw energy density at the epoch of its collapse is large enough , gws produced by the dw collapse may be observable in future experiments .
we have pointed out that such a mechanism is naturally built - in in the simplest model where the flaton couples to extra quarks as in ( [ w ] ) ; in such a model , the @xmath0 symmetry has an anomaly for su(3)@xmath6 and the qcd instanton effects serve as a bias for lifting the degeneracy among @xmath57 minima . in this case
the gw amplitude becomes so large that it can be well within the reach of future experiments .
k.n . would like to thank m. kawasaki and f. takahashi for useful conversations .
this work is supported by grant - in - aid for scientific research from the ministry of education , science , sports , and culture ( mext ) , japan , no .
22540263 ( t.m . ) , no .
22244021 ( t.m . ) , no .
21111006 ( k.n . ) , and no .
22244030 ( k.n . ) .
m. endo , k. hamaguchi , f. takahashi , phys .
lett . *
96 * , 211301 ( 2006 ) .
[ hep - ph/0602061 ] ; phys .
rev . * d74 * , 023531 ( 2006 ) .
[ hep - ph/0605091 ] ; s. nakamura , m. yamaguchi , phys .
* b638 * , 389 - 395 ( 2006 ) .
[ hep - ph/0602081 ] ; m. dine , r. kitano , a. morisse , y. shirman , phys . rev . * d73 * , 123518 ( 2006 ) . [ hep - ph/0604140 ] . | thermal inflation is an attractive solution to the cosmological moduli problem . however , domain walls may be formed after thermal inflation and some mechanisms are needed to eliminate the domain wall before it dominates the universe .
we point out that gravitational waves produced by the dynamics of domain walls may be observed by the pulsar timing experiments and future space - borne gravitational wave detectors , which provides a probe into the period of thermal inflation .
we also show that the qcd instanton effect can effectively eliminate the domain walls with producing observable amount of gravitational waves .
ut-11 - 20 + 1.0 cm |
recently , gill et al . introduce a new operator , the particle s position - momentum dot product @xmath8 , or _
posmom _ as they called , and establish a _ posmometry _ ( the distribution density of the posmom ) for some atomic and molecular systems .
posmometry1,posmometry2 the _ posmom _ operator in one of the cartesian axes , say , @xmath4 axis @xmath9 , is an essentially self - adjoint operator .
@xcite in ref .
[ ] , we developed this operator on the a two - dimensional spherical surface @xmath10 and successfully worked out its distribution densities for some molecular rotational states . in this letter , we explore the posmometry of quantum states on a circle which frequently model the planar rigid rotor , molecular rotation constrained on a plane , etc .. as embedding @xmath6 in the two - dimensional flat space @xmath7 , there are two operators @xmath11 @xmath12 that are respectively defined along two cartesian axes of coordinate respectively , which turn out to take following form , @xmath13 where @xmath14 , and @xmath15 , @xmath16 .
in fact , the momenta @xmath17 and @xmath18 are special case of the the so - called geometric momentum
@xmath19 @xcite on an @xmath20-dimensional surface which is embedded in @xmath21 dimensional euclidean space , where @xmath22 is the gradient operator on surface , and @xmath23 is the mean curvature and @xmath24 is the normal vector .
@xcite the explicit forms of @xmath25 and @xmath26 are,@xmath27 in the following section ii , i will present the elementary properties of this operator . in section iii
, i will give the posmometry for eigenstates of the @xmath1-component angular momentum @xmath28 , @xmath29 .
final section vi is our conclusions .
the following properties of the two posmom operators @xmath11 @xmath30 are easily attainable .
\i ) since the geometric momentum @xmath31 describes the motion constrained on the surface @xmath6 and there is no motion along the normal direction @xmath24 , which in quantum mechanics is expressed by @xmath32 while it is in classical mechanics expressed by @xmath33 .
this is why two operators @xmath25 and @xmath34 are linearly dependent , as shown in eq .
( [ indep ] ) . so it suffices to study one of the them , and i will concentrate @xmath25 .
\ii ) the operator @xmath25 has the reflection symmetry : @xmath35 in other words , posmom @xmath25 commutes with two parity operators @xmath2 and @xmath3 denoting two reflections about axes @xmath36
, @xmath37 respectively , and we have , @xmath38 = \left [ m_{y},q_{x}\right ] = 0.\ ] ] this mirror invariance is helpful in construction of the complete set of the eigenfunctions on the circle once the eigenfunctions of @xmath25 in four quadrants @xmath39 @xmath40 , @xmath41 and @xmath42 are known , respectively .
\iii ) in the full circle @xmath43 , we have the solution to the eigenvalue problem @xmath44 , @xmath45which is delta function normalized in any one of four quadrants , say in the first : @xmath46 . for convenience
, we can use @xmath47 ( @xmath48 ) to denote the eigenfunctions defined within the @xmath49th quadrants of the circle respectively .
\iv ) note that @xmath50 is not the simultaneous eigenfunction of operators @xmath51 , @xmath2 and @xmath3 . the complete simultaneous eigenfunction set of operators @xmath51 , @xmath2 and @xmath3 is given by , @xmath52 where @xmath4 and @xmath53 indicate even and odd parity respectively about @xmath4-axis , and so on , @xmath54 the orthonormality and completeness of the eigenfunction set ( [ parity1])-([parity4 ] ) are satisfied .
i.e. , we have following two relations , @xmath55 where @xmath56 standing for @xmath57 , and @xmath58 .
for any state @xmath59 on the circle , we have , @xmath60 where , @xmath61 .
this normalization clearly states that for a given @xmath62 , the distribution density @xmath63 usually comes form four parts , @xmath64 it means that for a given @xmath62 , the probability amplitude is from eq .
( [ complete2 ] ) a four - valued function , @xmath65 however , in the following section , we see that for the eigenstates @xmath66 of the @xmath1-component angular momentum @xmath67 , the probability amplitude is in general triple - valued .
as is well known , the @xmath1-component angular momentum @xmath69 has a complete set of eigenfunctions @xmath70 ( @xmath71 , @xmath72 , @xmath73 , ... ) that span a hilbert space for analyzing any state on @xmath6 .
in general , we have , @xmath74 where the expansion coefficients @xmath75 , @xmath76 , @xmath77 and @xmath78 are given by , @xmath79 with , @xmath80 in which with @xmath81 symbolizing the hypergeometric function , @xmath82because of the eigenfunctions @xmath83 can be decomposed into two parts according to mirror symmetry operators : @xmath84 , it is for our purpose sufficient to study the eigenfunctions @xmath83 with @xmath85 . evidently , for @xmath86 being a positive even number @xmath87 ( @xmath88 ) , we obtain , @xmath89 for @xmath86 being a positive odd number @xmath90 ( @xmath88 ) , we obtain , @xmath91 so , we see that the probability amplitude is in general triple - valued .
unfortunately , the expansion coefficients ( @xmath92 , @xmath93 , @xmath94 , @xmath95 ) , provided nontrivial , can not be all greatly simplified unless @xmath96 and @xmath86 being odd . for @xmath71 and @xmath87
, we have following relations , @xmath97 where @xmath71 is the only case the probability amplitude of the posmom is double valued . for @xmath86
being odd we have @xmath98 , and for @xmath99 , we have explicitly , @xmath100 the probability distributions @xmath101 ( [ p ] ) for rotational states represented by @xmath102 with @xmath71 to @xmath103 are plotted in fig . 1 ( @xmath104 ) , fig . 2 ( @xmath105 ) , fig .
3 ( @xmath106 ) , fig . 4 ( @xmath107 ) , fig . 6 ( @xmath108 ) and fig . 6 ( @xmath109 ) respectively . on the whole , they are similar to the momentum distributions of stationary states for the one - dimensional simple harmonic oscillator .
it is understandable from an examination of the free motion on the circle . in classical mechanics for the free motion with a frequency @xmath110
, we have @xmath111 and @xmath112 and therefore @xmath113 .
then , in a classical state , the posmom @xmath114 has a half period as @xmath4 or @xmath17 has . in classical limit , whenever @xmath86 being even or odd number , the eigenstate @xmath102 behaves like a simple harmonic oscillator as seen from the posmom . for the ground state @xmath115 ( solid line with highest peak near at @xmath116 ) , and that for 2nd excited state @xmath117 which is the sum of @xmath118 ( dotted ) and @xmath119 ( dashed ) .
neither has a node.,scaledwidth=90.0% ] for the 4th excited state @xmath120 , which is the sum of @xmath118 ( dotted ) and @xmath121 ( dashed ) but @xmath122 and @xmath121 differ appreciately only near @xmath123 .
this density has no node either.,scaledwidth=90.0% ] for the 6th excited state @xmath124 , which is the sum of @xmath118 ( dotted ) and @xmath121 ( dashed ) and we see again that @xmath118 and @xmath121 differ appreciately only near @xmath123 .
the density exhibits no node but two minimum points over interval of finite @xmath125 almost reach the zero.,scaledwidth=90.0% ] for the 1st ( dashed ) , 3rd ( dotted ) and 5th ( solid ) state @xmath126 , @xmath127 .
these distribution densities and the momentum distribution densities for the 0th , 1st and 2nd state of one - dimensional simple harmonic oscillator ( not shown ) , are respectively similar.,scaledwidth=90.0% ] figure 4 for the 40th excited state @xmath128 , which is symmetrical about @xmath129 but only portion over the postive @xmath62 is depicted .
it is the sum of @xmath130 ( dotted ) and @xmath131 ( dashed ) and they differ only near point @xmath129 ; both symmetrical about @xmath129 but half portions over the negative @xmath132 are plotted .
this distribution density has clearly @xmath133 peaks and @xmath134 minima in the interval of finite @xmath135 but apparently @xmath136 nodes . to note that two minima near @xmath129 appraoch closer as if they are a single one .
we can infer that in the limit of large @xmath86 that is even , it is more and more similar to the distribution density for the @xmath137th state of one - dimensional simple harmonic oscillator.,scaledwidth=90.0% ] figure 5 for the 41th excited state @xmath138 , which is the sum of two identical part @xmath130 and @xmath131 . the distribution density and the momentum distribution density for the @xmath134th excited state of one - dimensional simple harmonic oscillator ( not shown ) ,
are almost the same .
so , we can infer that in the limit of large @xmath86 that is odd , it becomes more and more similar to the distribution density for the @xmath139th state of one - dimensional simple harmonic oscillator.,scaledwidth=90.0% ]
the posmom offers a new way to understand the quantum motions , constrained or not .
this study explores the posmom on a circular circle , and identify that the momentum in it is the geometric momentum that is recently proposed to properly describe of the momentum for the motions constrained on the curved surface . for construction of the complete basis , we need to resort to the mutual commutativity between posmom and parity operators , and then obtain the satisfactory bases each of them is in general four - valued .
the posmometry of the eigenstates of the @xmath1-axis component of the angular momentum is worked out , and is found to be similar to the momentum distributions of stationary states for the one - dimensional simple harmonic oscillators .
then any states on the circle can thus go through the posmometry analysis .
once the posmometer is successfully designed and built up , the ground state of the planar rotation of some molecules , which can be easily prepared , can be visualized via the distribution of density of the posmom . | developing the analysis of the distribution of the particle s position - momentum dot product , the so - called _ posmom _ @xmath0 * * , * * to quantum states on a circular circle on two - dimensional cartesian coordinates , we give its posmometry ( introduced recently by y. a. bernard and p. m. w. gill , posmom : the unobserved observable , j. phys .
chem .
lett .
1**(**2010**)**1254 ) for eigenstates of the free motion on the circle , i.e. , @xmath1-axis component of the angular momentum .
the posmom has two parity symmetries , specifically , invariant under two operations @xmath2 and @xmath3 representing mirror symmetry about @xmath4 and @xmath5 axis respectively .
the complete eigenfunction set of the posmom is then four - valued and consists of four basic parts each of them is defined within a distinct quadrant of the circle .
the results are not only potentially experimentally testable , but also reflect a fact that the embedding of the circle @xmath6 in two - dimensional flat space @xmath7 is physically reasonable . |
by virtue of riemann mapping theorem for every pair @xmath0 of simply connected domains in the complex plane one can find a conformal mapping @xmath1 except for two cases : @xmath2 or @xmath3 .
the situation is quite different for doubly connected domains .
a domain @xmath4 is doubly connected if @xmath5 consists of two connected components ; that is , continua in the riemann sphere @xmath6 .
we say that @xmath7 is nondegenerate if both components contain more than one point .
every nondegenerate doubly connected domain can be conformally mapped onto an annulus @xmath8 where the ratio @xmath9 does not depend on the choice of the conformal mapping .
this gives rise to the notion of conformal modulus , @xmath10 in fact , modulus is the only conformal invariant for nondegenerate doubly connected domains .
let us set @xmath11 for the degenerate cases .
complex harmonic functions , whose real and imaginary parts need not be coupled by the cauchy - riemann system , provide significantly larger class of mappings , but still restrictions on the domains are necessary .
the studies of the harmonic mapping problem began with rad s theorem @xcite which states that there is no harmonic homeomorphism @xmath12 for any proper domain @xmath13 .
the first proof of rad s theorem in this form was given by bers @xcite ; both rad and bers were motivated by the celebrated bernstein s theorem : _ any global solution _ ( in the entire plane ) _ of the minimal surface equation is affine_. harmonic mapping problem for doubly connected domains originated from the work of j.c.c .
nitsche on minimal surfaces . in 1962
he formulated a conjecture @xcite which was recently proved by the present authors @xcite . a harmonic homeomorphism @xmath14 between circular annuli exists if and only if @xmath15 it is a simple matter to see that harmonic functions remain harmonic upon conformal change of the independent variable @xmath16 .
therefore , theorem a remains valid when the annulus @xmath17 is replaced by any doubly connected domain @xmath18 of the same modulus @xmath19 .
the nitsche bound then reads as @xmath20 the harmonicity of a mapping @xmath21 is also preserved under affine transformations of the target .
thus it is natural to investigate necessary and sufficient conditions for the existence of @xmath22 in terms of the conformal modulus of @xmath7 and an affine invariant of the target @xmath23 .
this leads us to the concept of affine modulus .
a @xmath24-affine automorphism of @xmath24 is a mapping of the form @xmath25 with @xmath26 , @xmath27 .
an @xmath28-affine automorphism of @xmath24 , or simply affine transformation , takes the form @xmath29 with determinant @xmath30 .
[ afmod ] the _ affine modulus _ of a doubly connected domain @xmath18 is defined by @xmath31 for an equivalent formulation and properties of the affine modulus see section [ affinesection ] .
we can now state our main results : a necessary condition ( theorem [ necessary ] ) and sufficient condition ( theorem [ sufficient ] ) for the existence of a harmonic homeomorphism @xmath1 .
[ necessary ] if @xmath32 is a harmonic homeomorphism between doubly connected domains , and @xmath7 is nondegenerate , then @xmath33 where @xmath34 is an increasing function such that @xmath35 as @xmath36 .
more specifically , @xmath37 [ gennit ] based upon the nitsche bound it seems reasonable to expect that @xmath38 but we have no proof of this . [ sufficient ] let @xmath7 and @xmath23 be doubly connected domains in @xmath24 such that @xmath39 then there exists a harmonic homeomorphism @xmath32 unless @xmath40 is bounded . in the latter case there is no such @xmath22 .
when @xmath41 , the comparison of inequalities and shows that both are asymptotically sharp .
we do not know if equality in ( with both sides finite ) would suffice as well .
this is discussed in remark [ equalmod ] .
we write @xmath42 when @xmath43 is contained in @xmath44 in such a way that the inclusion map @xmath45 is a homotopy equivalence .
for doubly connected domains this simply means that @xmath46 and @xmath47 .
the monotonicity of modulus can be expressed by saying that @xmath42 implies @xmath48 and @xmath49 .
observe that both conditions and are preserved if @xmath7 is replaced by a domain with a smaller conformal modulus , or @xmath23 is replaced by a domain with a greater affine modulus .
thus , theorems [ necessary ] and [ sufficient ] suggest the formulation of the following comparison principles . [
compared ] ( domain comparison principle ) let @xmath7 and @xmath23 be doubly connected domains such that @xmath7 is nondegenerate and there exists a harmonic homeomorphism @xmath50 . if @xmath51 , then there exists a harmonic homeomorphism @xmath52 .
[ comparet ] ( target comparison principle ) let @xmath7 and @xmath23 be doubly connected domains such that there exists a harmonic homeomorphism @xmath50 . if @xmath53 is nondegenerate and @xmath54 , then there exists a harmonic homeomorphism @xmath55 .
both problems [ compared ] and [ comparet ] are open to the best of our knowledge .
although the harmonic mapping problem has its roots in the theory of minimal surfaces , it also arises from the minimization of the dirichlet energy @xmath56= \iint_\omega { \lvertdf\rvert}^2\ ] ] among all homeomorphism @xmath57 .
the minimizers and stationary points of @xmath58 serve as a model of admissible deformations of a hyperelastic material with stored energy @xmath58 @xcite .
the existence of a harmonic homeomorphism @xmath50 is necessary for the minimum of @xmath59 $ ] to be attained in the class of homeomorphisms between @xmath7 and @xmath23 . indeed
, such minimal maps must be harmonic .
the rest of the paper is organized as follows : we discuss relevant affine invariants in section [ affinesection ] , prove theorem [ necessary ] in section [ necessarysection ] , and prove theorem [ sufficient ] in section [ sufficientsection ] .
we write @xmath60 , @xmath61 , and @xmath62 .
the harmonic mapping problem for multiply connected domains has been also studied in @xcite , where the authors mostly focus on existence of harmonic mappings onto some domain of a given canonical type ( such as disk with punctures ) rather than onto a specific domain .
in this section @xmath7 is a doubly connected domain in @xmath24 , possibly degenerate .
exactly one of the components of @xmath63 is bounded and is denoted @xmath64 .
we also write @xmath65 , which is a simply connected domain .
let @xmath66 denote the area ( planar lebesgue measure ) of a set @xmath67 .
we are now ready to introduce the first affine invariant of @xmath7 .
[ carmod ] the _ carleman modulus _ of @xmath7 is defined by the rule @xmath68 unless @xmath69 or @xmath70 , in which case @xmath71 .
the well - known inequality @xmath72 proved by t. carleman @xcite in 1918 , can be expressed by saying that among all doubly connected domains with given areas of @xmath64 and @xmath73 the maximum of conformal modulus is uniquely attained by the circular annulus @xmath17 , @xmath74 .
carleman s inequality was one of the earliest isoperimetric - type results in mathematical physics , many of which can be found in @xcite .
our second affine invariant arises from an energy minimization problem .
recall the capacity of @xmath7 , @xmath75 where the infimum is taken over all real - valued smooth functions on @xmath24 which assume precisely two values @xmath76 and @xmath77 on @xmath78 .
the conformal modulus , defined in , is given by @xmath79 since we are looking for an affine invariant of @xmath7 , the following variational problem is naturally introduced .
[ affmodulus ] define the _ affine capacity _ of @xmath7 by @xmath80 where the infimum is taken over all invertible matrices @xmath81 and over all real functions @xmath82 as in .
thus , the affine modulus of @xmath7 , defined in , is @xmath83 let us now examine the properties of the affine modulus .
from we see that @xmath84 when @xmath7 is a circular annulus , equality holds in and therefore in .
hence @xmath85 .
equality @xmath86 is also attained , for example , if @xmath7 is the _ teichmller ring _ @xmath87\cup [ s,+\infty)),\qquad s>0\ ] ] indeed , for any affine automorphism @xmath88 there is a @xmath24-affine automorphism @xmath89 that agrees with @xmath90 on @xmath28 . since @xmath91 and @xmath92 is conformal
, it follows that @xmath93 on the other extreme , @xmath94 may be infinite even when @xmath95 is finite . indeed , the domain @xmath96\ ] ] is affinely equivalent to @xmath97 $ ] for any @xmath98 .
the conformal modulus of the latter domain grows indefinitely as @xmath99 .
since the second inequality in becomes vacuous when @xmath69 or @xmath70 , it is desirable to have an upper estimate for @xmath100 in terms of other geometric properties of @xmath7 . recall that the _ width _ of a compact set @xmath67 , denoted @xmath101 , is the smallest distance between two parallel lines that enclose the set . for connected sets
this is also the length of the shortest @xmath77-dimensional projection .
[ finiteaffmod ] let @xmath7 be a nondegenerate doubly connected domain .
if @xmath102 , then @xmath103 let @xmath88 be an affine automorphism .
denote its lipschitz constant by @xmath104 . clearly @xmath105 and @xmath106
. now follows from the extremal property of the teichmller ring : it has the greatest conformal modulus among all domains with given diameter of the bounded component and given distance between components ( * ? ? ?
iii.a ) .
even when the affine modulus is finite , the supremum in is not always attained .
an example is given by the _ grtzsch ring _
@xmath107 indeed , @xmath108 by lemma [ finiteaffmod ] .
equality holds in because the images of @xmath109 under mappings of the form @xmath110 , @xmath111 , converge to the domain @xmath112\cup [ 2s,\infty))$ ] which is a @xmath24-affine image of @xmath113 .
yet , for any affine automorphism @xmath90 @xmath114\cup [ s,\infty))\}=\operatorname{mod}\mbox{${\mathcal{t}}\big(\frac{s-1}{2}\big)$}\ ] ] where the first part expresses the monotonicity of modulus , and the second follows from .
thus the supremum in is not attained by any @xmath90 .
[ equalmod ] since equality holds in , lemma [ finiteaffmod ] is sharp .
furthermore , the pair of domains @xmath115 and @xmath116 can serve as a test case for whether equality in implies the existence of a harmonic homeomorphism .
the identity @xmath117 somewhat resembles the relation between conformal moduli of the grtzsch and teichmller rings ( * ? ? ?
iii.a ) , @xmath118
before proceeding to the proof we recollect basic facts of potential theory in the plane which can be found in @xcite . a domain
@xmath7 has green s function @xmath119 whenever @xmath63 contains a nondegenerate continuum .
our normalization is @xmath120 as @xmath121 .
in particular , @xmath122 .
green s function for the unit disk is @xmath123 if @xmath124 is a holomorphic function , then the subordination principle holds : @xmath125 one consequence of and is a general version of the schwarz lemma . if @xmath126 is holomorphic and @xmath127 for some @xmath128 , then @xmath129 there is an application of to harmonic homeomorphisms
we refer to @xcite for a discussion of relation between harmonic and quasiconformal mappings , and to the book @xcite for the general theory of quasiconformal mappings .
[ distprop ] let @xmath4 be a domain with green s function @xmath119 .
fix @xmath130 . for any harmonic homeomorphism @xmath131
there exists an affine automorphism @xmath90 such that the composition @xmath132 satisfies @xmath133 for @xmath134 . replacing @xmath22 with @xmath135
, we may assume that @xmath22 is orientation preserving .
then @xmath22 satisfies the second beltrami equation @xmath136 where the _ second complex dilatation _ @xmath137 is an antiholomorphic function from @xmath7 into @xmath138 . since @xmath139 , the mapping @xmath22 is quasiconformal away from the boundary of @xmath7 .
affine transformations of @xmath22 do not affect harmonicity by may decrease @xmath140 .
indeed , the composition @xmath141 satisfies new second beltrami equation @xmath142 by setting @xmath143 we achieve that the second complex dilatation of @xmath144 vanishes at @xmath145 .
now the required estimate follows from .
we would like to make the estimate more explicit when the domain @xmath7 is a circular annulus @xmath146 .
although an explicit formula for green s function of an annulus can be found , for us it suffices to have a lower bound .
we obtain such a bound from the subordination principle .
indeed , for any @xmath147 the vertical strip @xmath148 has green s function @xcite @xmath149 we use with @xmath150 and @xmath151 where @xmath152 @xmath153 since @xmath154 maps @xmath155 onto @xmath146 with @xmath156 , inequality together with the subordination principle yield @xmath157 we are now ready to give an explicit estimate for holomorphic functions with at least one zero in an annulus .
[ slem ] let @xmath146 and suppose that @xmath126 is a holomorphic function with @xmath158 . then for each @xmath159 we have @xmath160 the case @xmath161 immediately follows from and . for the general case ,
let @xmath162 by hadamard s three circle theorem , @xmath163 is a convex function of @xmath164 .
since @xmath165 and @xmath166 for all @xmath167 , the convexity implies @xmath168 and the same estimate holds for @xmath169 . with the aid of conformal transformation of @xmath7 we may assume that @xmath146 where @xmath170 . as in the proof of proposition [ distprop ]
, we apply an affine transformation to obtain a harmonic mapping @xmath144 whose second complex dilatation vanishes at @xmath77 . by lemma [ slem ]
the restriction of @xmath144 to @xmath171 is @xmath172-quasiconformal ; that is , @xmath173 the conformal modulus of @xmath174 is distorted by the factor of at most @xmath172 under the mapping @xmath144 , see @xcite . hence @xmath175 we are free to choose any @xmath176 in .
introduce the function @xmath177 which is positive and decreasing for @xmath178 .
clearly @xmath179 as @xmath180
. now takes the form @xmath181 and @xmath35 as @xmath36 . to obtain a concrete bound
, we test the supremum in by putting @xmath182 obtaining @xmath183 which is .
for notational simplicity we denote by @xmath184 the class of harmonic homeomorphisms from @xmath7 onto @xmath23 .
this includes conformal mappings @xmath185 if they exist .
the proof of theorem [ sufficient ] is divided into three parts . in this section
we prove that @xmath184 is empty when @xmath186 and @xmath40 is bounded .
suppose to the contrary that @xmath187 .
up to a conformal transformation , @xmath7 is the circular annulus @xmath188 .
the target domain is @xmath189 for some compact set @xmath190 . with the help of inversion @xmath191
we can arrange so that @xmath192 approaches @xmath190 as @xmath193 . for large enough integers
@xmath194 the open disk @xmath195 contains @xmath190 .
let @xmath196 . as @xmath197 , the outer boundary of @xmath198 approaches @xmath199 .
thus there exist conformal mappings @xmath200 such that @xmath201 pointwise and @xmath202 as @xmath197 .
define a harmonic mapping @xmath203 and observe that for any @xmath204 the value @xmath205 is defined when @xmath194 is large enough .
moreover , @xmath206 because @xmath207 , and the convergence is uniform on compact subsets of @xmath208 , i.e. , @xmath209 let us write @xmath210 where the coefficients depend on @xmath194 as well but we suppress this dependence in the notation .
in fact , implies that for each fixed @xmath211 the coefficients @xmath212 tend to @xmath76 as @xmath197 . on the other hand
, @xmath213 extends to a sense - preserving homeomorphism @xmath214 , which leads to a contradiction with the following result .
[ weitsman ] ( weitsman @xcite ) let @xmath215 be a sense preserving homeomorphism with fourier coefficients @xmath216 , @xmath217
. then @xmath218 weitsman s inequality is sharp .
earlier hall @xcite proved with @xmath219 instead of @xmath220 .
the validity of with some absolute constant can be traced to the work of several authors , see @xcite .
it is worth mentioning that first arose as a special case of shapiro s conjecture ( * ? ? ?
* problem 5.41 ) , which posed that for any sense - preserving @xmath221-fold cover @xmath222 @xmath223 where @xmath224 depends only on @xmath221 .
this conjecture was proved by hall @xcite for @xmath225 and by sheil - small @xcite in general .
the rate of decay of @xmath226 remains unknown , see @xcite .
here we assume that @xmath227 but @xmath40 is unbounded . without loss of generality , @xmath228 where @xmath229 is a simply connected domain and @xmath230 .
for @xmath231 we define a mapping @xmath232 by @xmath233 note that @xmath234 for all @xmath235 .
the choice of @xmath236 is motivated by the fact that its inverse is harmonic : @xmath237 since @xmath238 maps the doubly connected domain @xmath239 onto @xmath23 , it remains to find @xmath240 such that @xmath241 .
the latter will follow from the intermediate value theorem once we prove @xmath242 as @xmath180 and @xmath243 as @xmath244 .
the latter is obvious , so we proceed to the proof of the former limit .
let @xmath245 .
the complement of @xmath239 has two components : one is the disk @xmath246 and the other contains a point with absolute value @xmath247 . by the extremal property of the grtzsch ring ( * ? ? ?
iii.a ) , @xmath248 does not exceed the conformal modulus of the grtzsch ring @xmath109 with @xmath249 as @xmath180 , we have @xmath250 and thus @xmath251 .
this completes the proof .
since harmonicity is invariant under affine transformations of the target , we may assume that @xmath252 .
there are two substantially different cases
. * case 1 . * the set @xmath253 is contained in a line .
thus , up to a @xmath24-affine automorphism , @xmath23 is the teichmller ring @xmath254\cup [ t,+\infty))\ ] ] we may and do replace @xmath7 with a conformally equivalent domain @xmath255 for some @xmath256 . thus our task is to harmonically map @xmath255 onto @xmath257 .
let @xmath258 be a number to be chosen later .
define a piecewise linear function @xmath259 by @xmath260 and consider the domain @xmath261 shown in figure 1 .
, height=153 ] let @xmath262 be a conformal mapping of the upper halfplane @xmath263 onto @xmath264 normalized by boundary conditions @xmath265 , @xmath266 , and @xmath267 .
it is important to notice that the boundary of @xmath264 satisfies the quasiarc condition uniformly with respect to @xmath268 ; that is , @xmath269 for any three points on @xmath270 such that @xmath271 separates @xmath272 and @xmath273 . by a theorem of ahlfors @xcite
@xmath274 extends to a @xmath172-quasiconformal mapping @xmath275 with @xmath172 independent of @xmath268 .
the latter can be expressed via the quasisymmetry condition ( see @xcite or ( * ? ? ?
11 ) ) : there is a homeomorphism @xmath276 , independent of @xmath268 , such that @xmath277 for all distinct points @xmath278 . applying to the triple @xmath279 , we find @xmath280 hence @xmath281 as @xmath282 . on the other hand , @xmath283
when @xmath284 since @xmath285 depends continuously on @xmath268 , there exists @xmath286 for which @xmath287 .
let us fix such @xmath268 . as observed above
, @xmath288 has a continuous extension to @xmath289 .
it takes the segments @xmath290 and @xmath291 into horizontal segments @xmath292 and @xmath293 respectively . by the reflection principle @xmath288 extends holomorphically across each segment , and we have @xmath294 .
it follows that the function @xmath295 extends harmonically to the entire teichmller ring @xmath255 .
consider the harmonic mapping @xmath296 which by construction is continuous in @xmath24 .
we claim that @xmath22 is a homeomorphism from @xmath255 onto @xmath257 .
since @xmath22 agrees with @xmath274 on @xmath28 , it follows that @xmath22 maps @xmath28 homeomorphically onto @xmath28 in such a way that @xmath297 , @xmath298 , and @xmath299 . to prove that @xmath22 is injective in @xmath255
, we only need to show that @xmath300 is a strictly increasing function of @xmath301 for any fixed @xmath302 .
the partial derivative @xmath303 is harmonic and nonconstant in @xmath304 , has nonnegative boundary values on @xmath28 , and is bounded at infinity .
thus @xmath305 in @xmath304 by the maximum principle @xcite and the claim is proved . *
* the set @xmath253 is not contained in a line .
we claim that there exists an affine automorphism @xmath90 such that @xmath306 .
if this holds , then @xmath307 can be mapped conformally onto @xmath308 and the composition with @xmath309 furnishes the desired harmonic homeomorphism . since @xmath252 and @xmath310 depends continuously on the coefficients of @xmath90 , it suffices to show @xmath311 let @xmath190 and @xmath312 be the bounded and unbounded components of the complement of @xmath23 , respectively .
we need a lemma .
[ trunk ] there exists a compact set @xmath313 such that the union @xmath314 is not contained in a line .
let @xmath315 for @xmath316 .
if @xmath190 is not contained in any line , then we can let @xmath317 be any nontrivial connected component of @xmath318 , with @xmath319 large .
if @xmath190 is contained in some line @xmath320 , then @xmath318 is not contained in @xmath320 when @xmath319 is large enough , and we can again choose @xmath317 to be one of its connected components .
there exist two parallel lines @xmath321 and @xmath322 such that any line between @xmath321 and @xmath322 meets both @xmath190 and @xmath317 .
we may assume that these lines are vertical , say @xmath323 and @xmath324 .
let @xmath325 . for each @xmath326
there is a segment of line @xmath327 that joins @xmath190 to @xmath312 and has length is at most @xmath328 .
the extremal length of the family of all such segments is at most @xmath329 , see ( * ? ? ?
i ) . under the affine transformation
@xmath330 the length of vertical segments is unchanged , while the width of their family becomes @xmath331 instead of @xmath145 .
thus the extremal length of all rectifiable curves connecting @xmath332 and @xmath333 tends to @xmath76 as @xmath334 .
the relation between modulus and extremal length ( * ? ? ? * ch .
i ) implies @xmath335 as @xmath334 .
this completes the proof of and of theorem [ sufficient ] .
since conjecture [ gennit ] is known to be true for circular annuli @xcite ( nitsche conjecture ) , it is natural to test it on other canonical doubly connected domains : the grtzsch and teichmller rings . precisely , the questions are as follows .
_ grtzsch nitsche problem _ : for which @xmath336 does there exist a harmonic homeomorphism @xmath337 ? _ teichmller nitsche problem _ : for which @xmath338 does there exist a harmonic homeomorphism @xmath339 ?
indeed , by virtue of theorem [ sufficient ] we only need to consider @xmath341 .
choose @xmath342 so that @xmath343 let @xmath344 $ ] .
choose a branch of @xmath345 in @xmath346 so that @xmath347 .
note that @xmath348 in @xmath346
. therefore , the harmonic function @xmath295 satisfies @xmath305 in @xmath346 .
it follows that @xmath349 is a harmonic homeomorphism of @xmath346 onto itself .
it remains to observe that @xmath22 maps the domain @xmath350 $ ] onto @xmath351 $ ] . | the harmonic mapping problem asks when there exists a harmonic homeomorphism between two given domains .
it arises in the theory of minimal surfaces and in calculus of variations , specifically in hyperelasticity theory .
we investigate this problem for doubly connected domains in the plane , where it already presents considerable challenge and leads to several interesting open questions . |
determining the phase structure of quantum chromodynamics ( qcd ) is one of the central problems in hadron physics . in particular , at low temperature ( @xmath0 ) and non - zero chemical potential ( @xmath2 ) , one may expect that quark matter has a fermi surface , which gives rise to an interesting complication ; if there is some attractive channel in the quark - quark interaction , the fermi surface becomes unstable and quarks form cooper pairs , leading to a color superconductivity ( cs ) with the color - gauge symmetry broken @xcite .
since the critical temperature ( @xmath1 ) of cs is expected to be small , it would be difficult to create the color - superconducting quark matter by heavy - ion collisions .
so it is natural that people have exclusively considered neutron stars as a physical system to which cs may have a relevance .
however , there is a possibility to obtain information on cs by heavy - ion collisions through possible _
fluctuations _ of the pair field .
in fact , it is known that in the ordinary superconductivity in metals , there exist fluctuations of cooper pairs even above the critical temperature where the pair condensate does not exist : such fluctuations affect various transport coefficients , which are experimentally observed@xcite . in the usual superconductivity in metals ,
the pairing interaction due to the phonon exchange is so weak that the mean field approximation works well except in the immediate vicinity of the critical point . in other words ,
the fluctuations become important only near the critical temperature . however , in the case of cs at intermediate densities , large fluctuations may survive even well above @xmath1 owing to the strong coupling which may be described by a low - energy effective lagrangian of qcd . if there are large fluctuations even well above @xmath1 , heavy - ion collisions on earth may have a possibility to be utilized to study the physics of cs through examining precursory phenomena of cs .
-@xmath2 plane in our model .
the solid and dashed line denote the critical line of a first and second order phase transition , respectively . ]
we use the two - flavor nambu - jona - lasinio model in the chiral limit , @xcite @xmath3 + g_{c}(\bar{\psi}i\gamma_{5}\tau_{2 } \lambda_{2}\psi^{c})(\bar{\psi}^{c}i\gamma_{5}\tau_{2}\lambda_{2}\psi),\end{aligned}\ ] ] where @xmath4 with @xmath5 being the charge conjugation operator , and @xmath6 and @xmath7 are the second component of the pauli and gell - mann matrices representing the flavor su(2 ) and color su(3 ) , respectively .
the phase structure of our model in the mean field approximation is shown in fig . 1 .
the dashed line denotes the critical line of the second order phase transition and the solid line first order phase transition .
the tricritical point where the order of the chiral transition changes from the second to first order is located at about @xmath8 mev .
now that the phase structure is determined , we investigate the @xmath0 dependence of the fluctuation of the diquark pair field in the wigner phase above @xmath1 ; we calculate the response of the pair field to an external one in the linear response theory . when an external field @xmath9 is added in the wigner phase , the induced pair field @xmath10 is given by @xmath11 where @xmath12 is a particle - particle correlation function corresponding diagrammatically to fig . 2 .
the function @xmath13 is still a function of imaginary time . by carrying out the analytic continuation ,
we then obtain , what we call , the response function @xmath14 . here
it should be noticed that the following equality holds @xmath15 on account of the self - consistency condition for the second - order phase transition of the diquark condensate at @xmath16 .
this equality is the origin to lead to the various precritical phenomena for csc .
the nature of the fluctuation to be discussed below will not be altered even when the phase transition is _ weak _ first order owing to the fluctuations of the gluon fields . with @xmath17,@xmath18,@xmath19 and @xmath20 at @xmath21mev ( a ) and @xmath22mev ( b ) .
, width=340 ] with @xmath17,@xmath18,@xmath19 and @xmath20 at @xmath21mev ( a ) and @xmath22mev ( b ) .
, width=207 ] the strength of the fluctuation is given by the spectral function @xmath23 where @xmath24 .
the @xmath0 dependence of the spectral function at @xmath25 mev and zero momentum transfer ( @xmath26 ) is shown in fig .
one can see a prominent peak moving toward the origin grows as @xmath0 is lowered toward @xmath1 as indicated by the arrow in fig . 1 . in the experimental point of view , it is interesting that the peak survives at @xmath0 even well above @xmath1 with @xmath27 .
this means that the precritical region of csc is one to two order larger in the unit of @xmath1 than that in the usual electric superconductors of metals .
the result for @xmath28 mev is shown in fig .
3(b ) ; one sees that although the growth of the peak becomes relatively moderate , the qualitative features do not change .
the existence of the peak with the narrow width suggests an existence of a well - defined collective mode as an elementary excitation in the wigner phase .
this can be examined by searching possible poles of the response function .
the existence of a pole @xmath29 with given @xmath30 means that the pair field @xmath31 can be created even with an infinitesimal external field @xmath9 as seen from eq .
( [ eqn : res]);i.e .
, the system admits spontaneous excitation of a collective mode with the dispersion relation @xmath29 .
( one should notice that the gap equation eq .
( [ eqn : critical ] ) implies that @xmath32 is a solution at @xmath16 , i.e. , there exists a zero mode at the critical point . ) the singular behavior seen in the spectral function near the critical point in fig .
3 is caused by the presence of the zero mode .
a numerical calculation shows that there indeed exists a pole in the lower half plane at @xmath33 .
4 shows how the pole moves as @xmath0 is lowered toward @xmath1 for @xmath25 and @xmath34 mev at @xmath26 .
one sees that the pole approaches the origin as @xmath0 is lowered toward @xmath1 .
such a mode whose energy tends to vanish as the system approaches the critical point of a phase transition is called a soft mode .
another characteristic feature of the soft mode is that the absolute value of the imaginary part @xmath35 of the pole is larger than that of the real part @xmath36 for both @xmath2 .
this feature may have an important implication for an effective equation describing the dynamical phase transition for csc : the dynamical behavior of the order parameter near @xmath1 is well described by a non - linear diffusion equation like the time - dependent ginzburg - landau equation .
such a soft mode gives rise to anomalous behaviors of several transport coefficients ; for example , an anomalous excess of the electric conductivity above @xmath1 which is known as paraconductivity ( pc)@xcite . the diagrammatic contributions to pc are known as aslamazov - larkin and maki - tompson terms , which are shown in fig .
it is worth emphasizing that these diagrams can be regarded as a modification of the self - energy of gauge fields , i.e. , the gluons and the photons , which decay into dilepton pairs .
this implies that photons and hence dileptons from the system can carry some information of the fluctuation of cs .
we also notice that the pairing fluctuations can give rise to a pseudo - gap phenomena known in high-@xmath1 superconductivity of copper - oxides
. it would be intriguing to pursue possible similarity of csc and high-@xmath1 superconductivity .
these are under investigations .
99 for recent reviews , see k. rajagopal and f. wilczek , chapter 35 in the festschrift in honor of b. l. ioffe , `` at the frontier of particle physics / handbook of qcd '' , m. shifman , ed .
, ( world scientific ) ; m. alford , ann .
nucl . part .
sci . * 51 * , 131 ( 2001 ) .
t. tsuneto , `` superconductivity and superfluidity '' , cambridge university press , 1998 .
m. kitazawa , t. koide , t. kunihiro and y. nemoto , phys .
* d65 * , 091504(r ) ( 2002 ) . l. g. aslamazov and a. i. larkin , sov .
solid state * 10 * , 875 ( 1968 ) ; k. maki , prog .
* 40 * , 193 ( 1968 ) ; r. s. thompson , phys .
rev . * b1 * , 327 ( 1970 ) . | we investigate possible precursory phenomena of color superconductivity at finite temperature @xmath0 with an effective theory of qcd .
it is found that the fluctuation of the diquark pair field exists with a prominent strength even well above the critical temperature @xmath1 .
we show that such a fluctuaiton forms a collective mode , the corresponding pole of which approaches the origin as @xmath0 is lowered to @xmath1 in the complex energy plane .
we discuss the possible relevance of the precursor to the observables to be detected in heavy - ion collisions . |
after a research effort that spans almost ten years , we have been able to derive exact solutions of the nonlinear differential equations that describe the equilibrium structures of differentially rotating , self gravitating fluids with cylindrical symmetry .
our results are directly applicable to quasi
equilibrium configurations that may develop early in the evolution of protostellar and protoplanetary disks , and we expect that they will help us understand the physical conditions that prevail in such systems long before the onset of accretion processes , gas ionization , and nonaxisymmetric evolution .
the nonrotating analogues of our equations have been legendary in the literature ; they are known as the nonlinear lane
emden , thomas fermi , and emden fowler equations .
introduction of rotation to these equations complicates matters considerably and very little analytical work has been carried out to date in this most interesting case .
our results are also relevant to a long standing problem in the formation of our solar system that has preoccupied professional astronomers and nonprofessionals for over two centuries , namely the locations of planetary orbits and the observed `` order '' in the present solar system .
naturally , we too have ended up considering the famous titius bode `` law '' of planetary distances but not in an effort to discover some hidden physical principle or previously unknown `` universal '' underpinning ; on the contrary , we saw this `` law '' as an adverse interpretation to different and more complex patterns that we detected in the actual data and that we describe in [ work ] .
we begin in
[ le][tb ] with a brief overview of the research that has been carried out by other researchers on these topics .
we also mention in [ work ] some of the history of our work that has finally tied up together all the pertinent issues .
we refer especially to our past unsuccessful attempts to resolve the question of planetary distances because it is such setbacks that eventually pave the way for a successful conclusion of the research effort .
emden differential equation ( lane 186970 ; emden 1907 ) describes the equilibria of nonrotating fluids in which internal pressure balances self - gravity .
spherically symmetric solutions of this equation came to the attention of astrophysicists when chandrasekhar included them in his 1939 monograph `` an introduction to the study of stellar structure , '' but interest in such solutions continued to be largely academic because real stars rotate and rotation destroys spherical symmetry and modifies their internal profiles and physical characteristics . in the latter half of the twentieth century ,
the isothermal solution , commonly referred to as the `` singular isothermal sphere , '' and its nonsingular modifications found some interesting applications to the structures of collisionless systems such as globular clusters and early - type galaxies ( binney & tremaine 1987 ; rix et al .
1997 ) ; to the structures of large
scale gaseous systems such as x ray halos around elliptical galaxies ( fabbiano 1989 ) and clusters of galaxies ( sarazin 1988 ; fabian 1991 ) ; and to simplified models of gravitational lenses ( kochanek 1995 ) .
emden s work also attracted the attention of physicists outside the field of astrophysics ( fowler 1914 ; thomas 1927 ; fermi 1927 ) who studied generalized polytropic forms of the lane
emden equation for specific polytropic indices @xmath3 .
the extensive studies of fowler ( 1914 , 1930 , 1931 ) produced some singular solutions for @xmath4 and established the so called emden
fowler equation in the literature , while the works of thomas ( 1927 ) and fermi ( 1927 ) produced the so called thomas fermi equation , an important milestone in atomic theory . at present , both of these equations continue to be the subjects of investigations by physicists and mathematicians alike .
physicists are drawn to the emden
fowler equation because it appears in the kinetics of landau ginzburg critical phenomena ( see the detailed account of dixon & tuszyski 1990 ) and in the kinetics of combustion ( frank kamenetskii 1955 ) .
mathematicians use these nonlinear equations as laboratories to study a wide variety of properties in their solutions positivity , uniqueness , singularities , monotonic vs. oscillatory behavior , and bifurcations usually without having analytic solutions at hand ( e.g. , wong 1975 ; lions 1982 ; see also goenner & havas [ 2000 ] for a comprehensive list of all known exact analytic solutions ; as well as berkovich [ 1997 ] and goenner [ 2001 ] for classifications of some solutions obtained from lie group symmetries ) . the cylindrical form of the classical lane
emden equation differs from the spherical form in just one coefficient and both forms have been included as special cases in studies of the generalized emden
fowler equation ( horedt 1986 ; dixon & tuszyski 1990 ; berkovich 1997 ; goenner & havas 2000 ) .
but until now , the cylindrical lane emden equation has not attracted attention on its own merit not even in astrophysics where it can be used as a simplified ( but nonlinear ) model of a disk like or cylindrical self gravitating gas .
the only analytic studies of the cylindrical form that we are aware of have been made by jeans ( 1914 ) and robe ( 1968 ) who found a bessel function solution of the linear @xmath5 case with uniform rotation and by stodlkiewicz ( 1963 ) and ostriker ( 1964 ) who solved the nonrotating isothermal ( @xmath6 ) case .
ostriker ( 1964 ) , in particular , also showed that the radius and the mass per unit length are finite in all models with @xmath7 and that , in contrast to the singular isothermal sphere , the nonrotating , nonsingular , isothermal cylinder has finite mass per unit length in spite of its infinite radial extent .
after the work of robe ( 1968 ) , some numerical studies of rotating isothermal cylinders and disks ( hansen et al .
1976 ; schmitz & ebert 1986 ; narita et al .
1990 ) and rotating polytropic cylinders ( schneider & schmitz 1995 ) also found nonmonotonic solutions for the equilibrium density akin to the bessel
function oscillatory solution of the @xmath8 polytrope in uniform rotation ; but understanding this behavior proved elusive and so the nonmonotonic density profiles were marginalized .
schneider & schmitz ( 1995 ) , in particular , adopted some differentially rotating profiles and obtained numerical solutions for the radial density profiles of models with negative polytropic indices ( @xmath9 ) .
these authors did not study the isothermal models of interest here , and they did not explain the unusual properties of their solutions
. nevertheless , their work is by far the most closely related study to our work and their polytropic results should be considered in conjunction with our results from isothermal models .
the numerical algorithm called the titius bode `` law '' has been known for 240 years ( e.g. , nieto 1972 ; lecar 1973 ; danby 1988 ) .
it relies on an ad hoc geometric progression to describe the positions of the planets in the solar system and works fairly well out to uranus but no farther ( jaki 1972 ) .
the same phenomenology has also been applied to the satellites of the gaseous giant planets ( neuh@xmath10user & feitzinger 1986 ) .
two modern brief reviews of the history along with criticisms of this `` law '' have been written by graner & dubrulle ( 1994 ) and by hayes & tremaine ( 1998 ) . currently , the general consensus is that a satisfactory physical basis has not been found for this numerical coincidence despite serious efforts by many researchers in over two centuries .
furthermore , opinions differ on whether such a physical basis exists at all . apparently , many researchers still believe that the titius bode algorithm does have a physical foundation and continue to work on this problem .
the last decade of the twentieth century , in particular , saw a resurgence of investigations targeting precisely two questions : the origin of the `` law '' ( graner & dubrulle 1994 ; dubrulle & graner 1994 ; li , zhang , & li 1995 ; nottale , schumacher , & gay 1997 ) and its statistical robustness against the null hypothesis ( hayes & tremaine 1998 ) .
hayes & tremaine ( 1998 ) found some statistical evidence that the titius
bode `` law '' could be related to the long term dynamical stability of the solar system .
their work is the latest in a long line of differing arguments made by several statisticians in the past and summarized by these authors ( see also lynch 2003 ) .
more importantly , hayes & tremaine ( 1998 ) dismissed without discussion all the previous purported explanations of the physical origin of the `` law '' as `` not entirely convincing . '' in our opinion , that is an example of good physical instincts , but it still requires some physical understanding of planetary distances before final judgment is passed .
we believe that the work presented below does provide the required understanding , and we will return to a broad discussion of the titius
bode `` law '' in [ dis ] below .
just as jeans ( 1914 ) had done years ago , we began studying rotating self gravitating cylinders in 1997 for the same reason ; as jeans put it : _ `` all the essential physical features of the natural three
dimensional problem appear to be reproduced in the simpler cylindrical problem , so that it seems legitimate to hope that an argument by analogy may not lead to entirely erroneous result[s ] . '' _
being unaware of the work published in french by robe ( 1968 ) , we solved analytically the @xmath8 polytropic lane
emden equation with cylindrical symmetry , uniform rotation , and proper boundary conditions ( christodoulou 1997 , unpublished ) .
( applying boundary conditions did not concern jeans who was interested in the stability of local deformations in compressible gases . )
the radial density profiles for @xmath8 and for sufficiently fast rotation exhibited permanent oscillations due to the zeroth
order bessel function @xmath11 that is the dominant part of the solution . numerical integrations for polytropes with @xmath12 also showed that the solutions for the density oscillate permanently with radius .
this generic oscillatory behavior then led us to consider the distribution of semimajor axes of planetary orbits in the solar system and their possible connection to the radial density peaks found in the equilibrium solutions .
cccccccccc + index & object & semimajor & titius bode & & inversion & & mean & & + @xmath13 & name & axis @xmath14 & distance & error & distance & error & distance & error & @xmath15 + & & ( au ) & ( au ) & ( % ) & ( au ) & ( % ) & ( au ) & ( % ) & + 1 & mercury & 0.387 & 0.4 & 3.4 & & & & & + 2 & venus & 0.723 & 0.7 & @xmath16 & 0.622 & @xmath17 & 0.694 & @xmath18 & 1 + 3 & earth & 1 & 1.0 & 0.0 & 1.050 & 5.0 & 1.124 & 12.4 & 2 + 4 & mars & 1.524 & 1.6 & 5.0 & 1.663 & 9.1 & 1.883 & 23.5 & 2 + 5 & ceres & 2.765 & 2.8 & 1.3 & 2.816 & 1.8 & 3.364 & 21.6 & 2 + 6 & jupiter & 5.203 & 5.2 & @xmath19 & 5.135 & @xmath20 & 6.151 & 18.2 & 2 + 7 & saturn & 9.537 & 10.0 & 4.9 & 9.992 & 4.8 & 12.20 & 27.9 & 2 + 8 & uranus & 19.19 & 19.6 & 2.1 & 16.93 & @xmath21 & 19.80 & 3.2 & 1 + 9 & neptune & 30.07 & 38.8 & 29.0 & 27.52 & @xmath22 & 29.34 & @xmath23 & 1 + 10 & pluto & 39.48 & 77.2 & 95.5 & & & & & + + + on closer inspection of the actual planetary data , we found patterns other than the titius
bode algorithm that could in principle also provide good fits to some sections of the data .
these patterns are summarized in table 1 where we list the observed semimajor axes @xmath14 of planetary orbits in our solar system ( e.g. kaufmann 1994 ) along with three empirical fits to the data , the titius
bode `` law , '' _ inversion _ , and an arithmetic mean .
the inversion distance for each orbit is the geometric mean of the actual semimajor axes of the two nearest neighbors .
the mean distance for each orbit is the arithmetic average of the actual semimajor axes of the two nearest neighbors .
relative errors are calculated for all three fits with respect to the observed values .
the inversion distances are listed in table 1 because this is the first pattern that we saw in the observed data rather than the titius
bode `` law . '' the mean distances are shown for comparison purposes ; they are very accurate only for those orbits in which either or both of the other two fits fail .
the surprisingly small errors in the arithmetic mean distances of venus , uranus , and neptune led us to the working hypothesis that the effect that may cause inversion in the intermediate orbits does not operate at small or at large distances ; moreover , it is smoothly replaced at the two ends by a slick new regularity that manifests itself as a modest arithmetic average . this smooth transition from inversion to arithmetic
averaging was another significant pattern that we saw in the data : for example , the orbit of uranus ( along with jupiter s orbit ) reproduces successfully the orbit of saturn by inversion while , on the other side , the saturn
uranus pair is clearly in arithmetic progression with neptune ; and the earth
venus pair shows inversion on the side of mars and arithmetic progression on the side of mercury .
we thought that this could not be a numerical coincidence because we found similar smooth transitions in the orbital distances of the regular ( and even some irregular ) satellites of the jovian planets . the geometric
mean spacing of the 110 au objects in table 1 implies that their orbits are _ inverted images _ of every other one with respect to the corresponding in between orbit ( e.g. , coxeter 1989 ) , i.e. , that the semimajor axes of any three consecutive orbits obey the relation @xmath24 in geometry , inversion is the gateway to hyperbolic space , where the inverted orbits would appear to be conveniently equidistant a perfect symmetry indeed , taking place in a space that we can not even visualize .
but we could not see how the orbital plane of the solar system could be so strongly curved in its middle and so flat at the two ends ; so eventually we abandoned this line of reasoning and the high degree of symmetry in hyperbolic space .
next we turned to geometric optics , the only part of physics where the same relation occurs .
( [ d1d3d2d2 ] ) can also be written as @xmath25 this is a _ mirror equation _ and implies that the @xmath26 and the @xmath27 orbits are mirror images of one another while the in between orbit plays the role of a concave mirror . also , the magnification ratio @xmath28 is clearly larger than 1 and approximately constant throughout the system of orbits in geometric progression ( i.e. , for @xmath29 ) ; and it reduces to @xmath30 for those orbits in arithmetic progression ( see table 1 ) .
this model appeared promising for a while because it suggested that some mechanism could potentially be responsible for tapering off the geometric progression to an arithmetic progression , as indicated by the magnifications of the orbits .
unfortunately , the mirror equation of geometric optics is not based on a fundamental principle per se , it is only a linear approximation valid for paraxial rays , and we did not think that its deeper underlying principle ( fermat s principle ) could be applicable to planetary orbits . because a mirror equation similar to eq .
( [ mirror ] ) is not derived from first principles in any other part of physical theory , we finally became convinced that the observed planetary distances , the titius bode algorithm , and the inversion distances could not be derived from a physical principle ; and we redirected our effort toward mathematical reasons that might be responsible for the observed order in the solar system .
this prompted us to return to the lane
emden equation and to focus exclusively on its intrinsic properties in the cases with and without rotation . in the case of no rotation or for some specific rotation profiles , the lane
emden equation is scale invariant and can be transformed to an autonomous differential equation ( see e.g. bender & orzag [ 1978 ] and visser & yunes [ 2003 ] ) for the theory and the transformations of such differential equations ) . scale invariance was also exploited by graner & dubrulle ( 1994 ) and dubrulle & graner ( 1994 ) who argued that cold , self gravitating , perfect
fluid disks are scale invariant and that this condition is sufficient to generate unstable radial modes that are equidistant in @xmath31 ( see also schmitz [ 1984 ] and li et al .
[ 1995 ] for stability analyses that have effectively led to the same result ) .
unfortunately , the scale invariance of the inviscid hydrodynamical equations is easily broken by the chosen boundary conditions for the equilibrium system _ and _ for the stability problem ; because of this fact , we thought that we should not look for a geometric progression of the titius bode kind in the stability problem .
after all , scale invariant unstable modes , such as those studied by dubrulle & graner ( 1994 ) , could only produce a generic geometric progression and they would be incapable of matching the observed loss of inversion at small and large distances , where planetary orbits seem to taper off neatly to two different arithmetic progressions ( see table 1 ) . therefore , a mechanism based on such modes of disturbance would suffer from the same problem that also afflicts inversion to a hyperbolic space , the titius bode geometric progression , and any other idea that overemphasizes the geometric spacing of the intermediate planetary orbits and ignores the observed arithmetic progressions of orbits in the inner and the outer solar system .
returning to the scale invariance of the lane
emden equation itself , we realized that it would not matter if this equation lost this symmetry by the applicable boundary conditions so long as its autonomous form were to exhibit _ discrete scale invariance _
( dsi ; sornette 1998 ) . in the theory of autonomous systems , dsi is a stronger constraint than scale invariance , and it is associated with _
limit cycles _ in the phase portrait of the differential equation ( see appendix b in visser & yunes [ 2003 ] ) .
a solution that exhibits dsi is not invariant for any arbitrary rescaling of the independent variable , but it still is self similar for a specific rescaling of that variable .
this fact led sornette ( 1998 ) to propose that the titius bode `` law '' may be a manifestation of dsi in the solar system , where the constant ratio of the titius bode geometric progression or , equivalently , the constant magnification ratio @xmath32 of inversion is the specific factor that rescales each planetary orbit to the next outward orbit .
we followed then the hypothesis of sornette ( 1998 ) and the methodology of visser & yunes ( 2003 ) and we constructed numerically the phase portraits of the autonomous forms of the polytropic and isothermal lane
emden equations , where we looked for limit cycles and discrete self similar behavior with absolutely no success .
at that point , being convinced that there is no fundamental physics behind the observed patterns shown in table 1 above , and with the theory of nonlinear autonomous systems revealing no intrinsic periodicities , we basically had only one remaining option : to return to our initial approach and solve the full boundary value problem for the differentially rotating equilibrium systems in order to see if any interesting patterns would emerge from the differentially rotating solutions patterns similar to the bessel
function oscillations found in the uniformly rotating @xmath8 polytropic case that did not exist intrinsically in the differential equation but were generated and governed by the applied boundary conditions .
we proceeded to do just that , and our results for the isothermal case are described in the sections that follow below . in the end , we were truly surprised by the simplicity of the mechanism that generates the patterns seen in table 1 ; because the discord between the equilibrium density profile favored inherently by the differential rotation and the profile imposed externally by the boundary conditions , although low key and inconspicuous , it is nonetheless a plain fact and does provide a simple resolution to the long standing , centuries old problem of planetary `` order '' in our solar system ( see [ compo ] ) .
the remainder of the paper is organized as follows . in [ models ] , we solve analytically for the equilibrium structure of the midplane of a gaseous isothermal disk , incorporating in the lane emden equation the effects of self - gravity , differential rotation , and thermal pressure . in
[ solar ] , we adopt a four parameter analytic solution as our baseline and we use the rotation profile of the baseline to compute numerically the corresponding oscillatory equilibrium solution that obeys physical boundary conditions at the center .
then we fit the density maxima of this solution to the planetary orbits in the present solar system in order to determine the underlying physical characteristics and the stability properties of the baseline model . in
[ discussion ] , we discuss the significance of our results for our solar system and for protoplanetary disks in general .
we consider the axisymmetric equilibria that are available to a rotating self gravitating gas in the absence of viscosity and magnetic fields .
we adopt cylindrical coordinates ( @xmath33 ) and the assumption of cylindrical symmetry ( @xmath34 ) which lets us ignore @xmath35dependent gradients and reduces the problem to one dimension , the distance @xmath36 from the rotation axis .
this technique has become common practice in studies of rotating , self gravitating , fluid disks ( e.g. , goodman & narayan 1988 ; christodoulou & narayan 1992 ; christodoulou 1993 ; christodoulou , contopoulos , & kazanas 1996 ) because it simplifies the stability analyses of effectively two dimensional modes of disturbance . in what follows ,
we are interested in equilibrium structures that describe the physical conditions across the midplane of a protoplanetary disk , so the assumptions @xmath37 allow us to tackle the problem by solving ordinary differential equations ( odes ) .
we further adopt a rotation law of the form @xmath38 where @xmath39 is a dimensionless radius and the length scale @xmath40 will be specified in eq .
( [ length ] ) below .
furthermore , @xmath41 is the angular velocity , @xmath42 is the value of @xmath43 at some fixed radius , and the dimensionless function @xmath44 for differential rotation is generally an arbitrary function of @xmath45 . for centrally condensed models ,
it is convenient to choose @xmath46 and the regularity condition @xmath47 , while for singular or annular models , we choose @xmath48 and the normalization @xmath49 .
finally , we assume an isothermal equation of state of the form @xmath50 where @xmath51 is the thermal pressure , @xmath52 is the gas density , and @xmath53 is the constant isothermal sound speed .
additional calculations in which we used polytropes with indices @xmath54 and the results of schneider & schmitz ( 1995 ) who used polytropes with indices @xmath9 demonstrate that the characteristics of the solutions presented below are largely insensitive to the particular choice of @xmath3 .
axisymmetric and cylindrically symmetric , nonmagnetic equilibria for a perfect fluid are described by the equation @xmath55 where the gravitational potential @xmath56 satisfies poisson s equation @xmath57 where @xmath58 is the gravitational constant .
combining eqs .
( [ rot])([poisson ] ) and using the definition @xmath39 , we find a second order nonlinear innonhomogeneous ode that can be cast in the form @xmath59 where @xmath60 , @xmath61 is the maximum density or a fixed cutoff density for singular models , @xmath62 , @xmath63 , and @xmath64 the term @xmath65 represents the gravitational ( jeans ) frequency and the dimensionless rotation parameter @xmath66 measures centrifugal support against self gravity ; in general , @xmath67 , since the gas is also partially supported by pressure gradients in the radial direction .
( [ main1 ] ) reduces to the classical isothermal lane
emden equation in the absence of rotation ( @xmath68 ) .. this singular rotation law is not of interest in this work . ] we derive analytically the nonrotating solutions in [ norot ] using the modern theory of nonlinear odes ( e.g. , bender & orzag [ 1978 ] ;
see also the classic works of stodlkiewicz [ 1963 ] and ostriker [ 1964 ] ) .
then , in [ withrot ] , we derive analytically a class of particular solutions of the full problem ( eq . [ [ main1 ] ] with arbitrary @xmath44 differential rotation ) and , in [ compo ] , we discuss a subset of composite power
law solutions that are astrophysically interesting despite the fact that they are incapable of obeying the proper boundary conditions at @xmath69 . in the absence of rotation ,
the isothermal lane - emden equation ( eq . [ [ main1 ] ] with @xmath68 ) reads @xmath70 this equation is scale invariant under the transformation ( @xmath71 , @xmath72 ) , where @xmath73 is an arbitrary constant and @xmath74 .
therefore , it can be transformed to an autonomous form : using this value of @xmath75 , we define @xmath76 and we cast eq .
( [ main2 ] ) into an equidimensional in@xmath45 equation for the new function @xmath77 .
this equation is invariant under the transformation @xmath71 ( i.e. , its @xmath78 ) and reads @xmath79 finally , we let @xmath80 and we euler transform the independent variable ( @xmath81 ) to obtain the autonomous form for the function @xmath82 : @xmath83 where the dots denote derivatives with respect to @xmath84 . since the first derivative is missing from eq .
( [ main4 ] ) , we can integrate twice .
the first integral is @xmath85 and the solution is @xmath86 where @xmath87 and @xmath88 are integration constants .
the integral in eq .
( [ int2 ] ) can be calculated in closed form @xmath89 and a series of backsubstitutions produces the following general solution for the density @xmath90 : @xmath91 where @xmath92 and @xmath93 are arbitrary _ positive _ constants .
the condition that @xmath94 is physical and ensures that the density profiles are nonnegative . the condition that @xmath95 is not a limitation : eq .
( [ sol1 ] ) is invariant under the transformation ( @xmath96 , @xmath97 ) because @xmath93 contains implicitly the @xmath98 duality seen in eq .
( [ int2 ] ) above .
therefore , only positive values of @xmath93 need to be considered without this causing loss of generality . the equilibrium solutions obtained in eq .
( [ sol1 ] ) can be classified into three types : 1 . for @xmath99 ,
the density profiles are singular at @xmath69 and decrease monotonically for @xmath100 .
these solutions are analogues of the singular isothermal sphere .
2 . for @xmath101 ,
the density profile is centrally condensed and can easily satisfy proper boundary conditions at @xmath69 . using @xmath102 , eq .
( [ sol1 ] ) reduces to the stodlkiewicz
ostriker solution @xmath103 for which @xmath104 and @xmath105 .
3 . for @xmath106
, each density profile has a hole at @xmath69i.e .
, @xmath107and peaks at a finite radius @xmath108 , namely @xmath109 where @xmath110 and @xmath111 ; this normalization fixes the value of @xmath92 for any choice of @xmath106 .
the two constants are related by the equation @xmath112 the mass in all of these models is strongly concentrated around the density maximum @xmath113 ; this gives them the appearance of slender annuli despite the presence of extended regions of very low densities on either side of the maximum ; regions that extend all the way to @xmath69 and to @xmath114 .
also note that the solutions for @xmath115 are concave and shallow near the center where they have , not only @xmath107 , but also @xmath105 ; while the solutions for @xmath116 are convex and steeply rising near the center where they have @xmath117 as @xmath118 .
at large radii ( @xmath119 ) , all solutions are decreasing with radius and the density falls off as @xmath120 .
this rapid asymptotic decline , which is steeper than @xmath121 , is responsible for keeping the mass per unit length @xmath122 finite in all models . a direct integration of eq .
( [ sol1 ] ) shows that @xmath123 , independent of the constant @xmath92 . letting @xmath101 in this equation , we recover ostriker s ( 1964 ) result , @xmath124 , for the centrally condensed cylinder .
the isothermal lane - emden equation with differential rotation ( eq . [ [ main1 ] ] ) is repeated here for the purpose of discussion : @xmath125 when the right hand side ( hereafter rhs ) of this equation is nonzero ( i.e. , when @xmath126 and @xmath127 ) , the property of scale invariance is lost from all cases of interest ( uniform rotation , power law rotation , etc . ) , irrespective of the prescription chosen for the differential rotation function @xmath44 . ) with a nonzero rhs is scale invariant only for @xmath128 , where @xmath92 and @xmath129 are arbitrary constants .
this case can be solved by transforming the scale
invariant ode to its autonomous form ( as was explained in [ norot ] ) , but there is no need to do so ; its solution is obtained easier by the method described in this subsection . ] eq .
( [ main3 ] ) has no special symmetry associated with it , and this is enough to make many researchers turn the other way .
this is probably why some interesting features of eq .
( [ main3 ] ) that we describe below have gone unnoticed for so long . the rhs of eq .
( [ main3 ] ) is not merely a rotation
dependent correction term to the classical isothermal lane
emden equation ( [ main2 ] ) .
the introduction of rotation changes the properties of the ode to such a large extent that the nonrotating solutions found in [ norot ] can not guide the effort to find rotating equilibrium solutions .
in fact , it is the functional form of the rhs that determines now the structure of the solutions of the entire ode : by inspection of eq .
( [ main3 ] ) , we can write down an entire class of particular equilibrium solutions , namely @xmath130 provided that @xmath131 also holds true . using eq .
( [ part1 ] ) , we write @xmath132 and substituting this form into eq .
( [ part2 ] ) we find an ode for all the differential
rotation laws @xmath44 that satisfy eq .
( [ part2 ] ) identically and make eq .
( [ part1 ] ) a family of exact solutions of the lane
emden equation with rotation ( eq . [ [ main3 ] ] ) : @xmath133 this third order linear ode can be integrated directly to yield the following results : @xmath134 implying that @xmath135 and @xmath136 where @xmath92 , @xmath129 , and @xmath93 are arbitrary integration constants ( that are unrelated to those used in [ norot ] ) and @xmath137 implying that @xmath138 for all values of @xmath93 . with so many free parameters ( @xmath92 , @xmath129 , and @xmath93 ) in the differential rotation profile , these results can easily become a theorist s playground . here
we highlight just a few interesting points : 1 . _
parameter constraints_.eq .
( [ class2 ] ) shows that @xmath139 only for @xmath94 .
this constraint also limits the physical values of @xmath93 when @xmath140 in eq .
( [ class3 ] ) ; for example , @xmath141 when @xmath142 .
this limitation can be easily circumvented by implementing composite rotation profiles with @xmath143 ( see item 4 in this list and [ compo ] below ) .
2 . _ monotonically decreasing profiles_.eq .
( [ class2 ] ) shows that @xmath90 is a decreasing function of @xmath45 for @xmath144 .
the same condition is sufficient to also make @xmath44 a decreasing function of @xmath45 provided that @xmath145 in eq .
( [ class3 ] ) .
_ uniform rotation_.for @xmath146 , @xmath142 , and @xmath147 , eq .
( [ class3 ] ) reduces to @xmath148 and the equilibrium density ( eq . [ [ class2 ] ] ) then is @xmath149 . note that this constant can not be adjusted freely , e.g. , it can not be reset to 1 ; the requirement that @xmath148 fixes @xmath92 in eq .
( [ class3 ] ) and then the uniform density gets fixed to @xmath150 by eq .
( [ class2 ] ) .
4 . _ composite profiles with _ @xmath143.steep density profiles with @xmath151 can be obtained by selecting @xmath152 and by incorporating a central region of uniform rotation ( see [ compo ] for details ) .
even more complex equilibrium profiles can be constructed by combining two or more power laws .
a composite profile is demonstrated in [ solar1 ] below , where we connect two disjoint regions of constant density with a power law and
we apply the result to the early structure of the solar nebula .
asymptotic regime_.for @xmath151 and @xmath143 , eq.([class4 ] ) shows that @xmath153 as @xmath154 and eq.([class3 ] ) then exhibits the asymptotic behavior @xmath155 .
therefore , all steep density profiles with @xmath151 and @xmath156 approach a flat rotation curve ( @xmath157 constant ) at large radii , independent of the value of @xmath93 . from the perspective of the physics that dictates the above profiles , the solutions ( [ part1 ] ) of the lane
emden equation ( [ main3 ] ) describe a class of rotating self gravitating equilibria in which @xmath35gradients are neglected and the _ radial gradient _ of the gravitational acceleration is balanced exactly by the _
radial gradient _ of the centrifugal acceleration at every radius @xmath36 .
this occurs because , in the lane
emden equation , we have gone to second order by taking an extra derivative on the components of the equation of hydrostatic equilibrium .
the balance of gradients can be seen , most easily , by substituting eq .
( [ part1 ] ) into the one dimensional poisson s equation @xmath158 , where @xmath159 is the normalized potential ; the result is @xmath160 \ = \
\ \frac{1}{x}\frac{d}{dx } x \left[\frac{1}{2}\beta_0 ^ 2 \cdot x f^2\right ] \ , .
\label{gradients}\ ] ] in this equation , the bracketed terms are the gravitational and centrifugal accelerations , respectively .
this type of balance is different than the balance commonly discussed between the magnitudes of these two accelerations in rotating gravitating systems ; and the power
law density solutions are borne out of this conformance of the two gradients , whereas the familiar stalemate between centrifugal and gravitational force is only relevant to purely homogeneous fluid equilibrium systems or particle systems with no pressure support . in the isothermal gaseous case of interest here , a power
law density profile satisfies naturally the condition that the _ radial variation _ of the enthalpy gradient @xmath161 be zero ( see eq . [ [ part2 ] ] ) and so the pure power
law profile is not at all influenced by the radial variation of the pressure gradient it is an exact solution of eq .
( [ main3 ] ) . ) , namely @xmath162 is not satisfied by pure power
law profiles ( except in the trivial case with @xmath163 ) ; then , pressure gradient variations do affect the structure of the underlying equilibrium solutions , but not in a dramatic fashion . ] many of the rotation profiles discussed above are singular at @xmath69 . the solutions for the density , especially , are all pure power laws and , for @xmath164 , they all diverge as @xmath118 . as was mentioned in [ withrot ] , this is not a serious problem because the singularity at the center can be removed by assuming that the central region rotates uniformly and that the density profile switches to a power law beyond a `` core '' radius @xmath165 .
the core radius @xmath166 can be chosen freely even for steep density profiles with @xmath151 , but then the constant @xmath129 in eq .
( [ class3 ] ) must be positive and it should be adjusted accordingly so that @xmath44 is everywhere positive and monotonically decreasing with @xmath45 .
it turns out that the physical requirement that @xmath167 is weaker than the monotonicity condition that @xmath168 . with @xmath169 to ensure that @xmath139 and assuming that @xmath152 and @xmath151 , we find from eq .
( [ class3 ] ) that @xmath167 if @xmath170 where @xmath171 ; and the stronger condition that @xmath168 if @xmath172 therefore , all composite models with a uniformly rotating core region @xmath173 must satisfy the stronger condition ( [ cond2 ] ) for any choice of @xmath169 and @xmath151 in their equilibrium density profiles ( eq . [ [ class2 ] ] ) .
the uniformly rotating cores of the composite models discussed above call attention to another interesting feature : the density in these models must be constant and equal to @xmath150 in order to support this type of rotation ( see also item 3 in the list of [ withrot ] ) .
so these models can not obey the boundary condition that @xmath104 for centrally condensed structures .
more generally , all the power law solutions that we derived for the density in [ withrot ] and all the composite models , although they are exact intrinsic solutions of the lane
emden equation with rotation ( eq . [ [ main3 ] ] ) , they do _ not _ solve the associated boundary value problem .
the question then is : how are the actual density profiles of centrally condensed equilibrium models going to behave once the proper set of boundary conditions \{@xmath174 , @xmath175 } are imposed at the center ?
the answer can be obtained by numerical integrations that enforce the desired central boundary conditions in eq .
( [ main3 ] ) and in the analogous polytropic lane emden equation with rotation ; and by examining the analytic solution to the full boundary value problem of the linear polytropic case with index @xmath8 and uniform rotation .
all the numerical ( isothermal and polytropic ) solutions , as well as the @xmath8 analytic solution , , the solution to the boundary value problem \{@xmath174 , @xmath175 } is analytic ( robe 1968 ) : @xmath176 where @xmath177 is the zeroth order bessel function of the first kind and all the other symbols are defined as in this work .
( however , for @xmath8 polytropes , @xmath178 , where @xmath179 is the polytropic constant , and @xmath180 . )
it is clear in this solution that the bessel function oscillates permanently about the @xmath181 line , which is a particular solution of the @xmath8 linear ode . ]
demonstrate routinely ( see also schneider & schmitz [ 1995 ] ) that the equilibrium density profiles lose their monotonicity the first time they cross below the corresponding particular solutions ( solutions analogous to those discussed in [ withrot ] for the isothermal models ) . once the first such crossing occurs at some relatively small radius ( see figs .
[ fig2 ] and [ fig3 ] below ) , the actual physical solutions recognize the existence of the corresponding intrinsic solutions and they turn and oscillate permanently about the density level defined by these particular solutions .
this of course happens because the particular solutions are fundamental `` baseline '' solutions of the ode itself , regardless of externally imposed conditions .
when a set of external conditions are imposed at @xmath69 for physical reasons , the actual solutions emerging from the center do not match the baseline solutions ( since the power law behavior of the baseline is incompatible with the imposed conditions ) , but they are nonetheless attracted to them because the baseline solutions satisfy the ode inherently .
the result then is a permanent mismatch around the baseline that extends over all radii .
this behavior is demonstrated in [ solar2 ] below , in the example model shown in fig .
[ fig2 ] and , notably , in the composite isothermal model of the solar nebula shown in fig .
the oscillatory behavior of the density profiles discussed in [ compo ] finds a natural application to the structure of the midplane of the solar nebula . for this application
, we need a composite model as a baseline because such models allow for equilibrium density power laws that can be arbitrarily steep . to this end
, we formulate in [ solar1 ] a composite model using the exact analytic solutions determined in
[ withrot ] above .
then , in [ solar2 ] , we solve numerically the corresponding boundary value problem that exhibits the same rotation profile as the analytic model ; and we obtain an oscillatory solution for the density profile subject to the applied physical boundary conditions .
finally , we proceed to fit the density peaks of this model to the observed planetary distances in our solar system and we conclude in [ solar3 ] by determining important physical parameters associated with the structure , the dynamics , and the stability of the solar nebula . for our baseline equilibrium model of the midplane of the solar nebula , we adopt a composite analytic solution in which the isothermal density profile has the form of a truncated power law : @xmath182 where @xmath166 is the radius of the constant density core region , @xmath183 is the truncation radius beyond which the density remains constant at a low value , and the power
law index @xmath184 is defined by @xmath185 the condition that @xmath186 implies that @xmath187 and excludes the logarithmic rotation laws from consideration : for @xmath187 , logarithms are not introduced in the general form of the rotation law ( eq . [ [ class3 ] ] ) by the lower branch of eq .
( [ class4 ] ) .
moreover , we consider below an even more limited range of indices , namely @xmath188 , since we are primarily interested in steep density profiles with @xmath151 .
introducing @xmath166 and @xmath183 in the above profile is equivalent to specifying three different values for the constant @xmath92 in eq .
( [ class2 ] ) of [ withrot ] .
these values are chosen so that the density profile remains continuous as it switches from one branch to the next , namely @xmath189 the rotation law can then be determined from eqs .
( [ class3 ] ) and ( [ den30 ] ) by finding the values of the constant @xmath129 that also make this profile continuous at the junctions where @xmath165 and @xmath190 , namely @xmath191 x_2^{\textstyle 2 } \ , & \ { \rm if } \ \ \ x \geq x_2 \end{array } \right .
\label{den300}\ ] ] as @xmath192 from the right , the value of @xmath129 in the intermediate branch satisfies marginally the monotonicity requirement ( eq . [ [ cond2 ] ] )
determined in [ compo ] above . in practice
, it is easier to integrate the differential equation ( [ class1 ] ) in each of the three regions of the model and use the integration constants along with eq .
( [ den30 ] ) to ensure continuity at the junctions @xmath166 and @xmath183 . using either method
, we find that the rotation law has the form @xmath193 } \ , \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ { \rm if } \ \ \
x_1 < x < x_2 \\ \\ \sqrt{\frac{\textstyle\delta}{\textstyle\delta - 2 } \left [ \left(x_1/x_2\right)^{\textstyle 2 } - \left(x_1/x_2\right)^{\textstyle\delta } \right ] \left(x_2/x\right)^{\textstyle 2 } + \left(x_1/x_2\right)^{\textstyle\delta } } \ , \ \ \ \ \ \ \ \ { \rm if } \ \ \ x \geq x_2 & \end{array } \right . .
\label{den31}\ ] ] it is easy to show that this rotation law obeys the physical requirements that @xmath194 and @xmath168 at all radii for any choice of the parameter set \{@xmath195 , @xmath196 , @xmath188}. notice that , outside the core , the rotation profile is monotonically decreasing everywhere ; and that it becomes asymptotically flat at very large radii : as @xmath154 , then @xmath197 thus , in contrast to the pure power law density profiles ( item 5 in [ withrot ] ) , this composite profile exhibits nearly uniform rotation at very large distances .
this is because the density of the model is not allowed to decrease at large distances ; instead , it is kept constant at the low level shown in eq .
( [ den3 ] ) for @xmath198 .
the differential rotation function @xmath44 of eq .
( [ den31 ] ) and the corresponding equilibrium density profile @xmath199 of eq .
( [ den3 ] ) are shown in figure 1 for @xmath200 , @xmath201 , and for various choices of the power
law index @xmath188 .
the above composite equilibrium model is characterized by four free parameters : the core radius @xmath166 , the truncation radius @xmath183 , the rotation parameter @xmath202 , and the power
law index @xmath188 .
the density profile ( eq . [ [ den3 ] ] ) of this baseline solution of the lane emden ode ( eq . [ [ main3 ] ] ) is not capable of satisfying physical boundary conditions at the center and it serves only as a mean approximation to the density of the corresponding physical model .
the general form of the rotation law of the baseline ( eq . [ [ den31 ] ] ) can , however , be adopted for the differential rotation @xmath44 of the physical model as well .
then eq .
( [ den3 ] ) provides a prescription for the rhs of the lane
emden equation ( [ main3 ] ) which can thus be written as @xmath203 this ode can be integrated numerically subject to the physical boundary conditions that @xmath204 an example is shown in figure 2 for an equilibrium model with @xmath205 , @xmath206 , @xmath207 , and @xmath208 .
notice that , in the linear scale of fig .
2a , the actual density peaks ( along the solid line ) are approximately equidistant in regions where the baseline density ( dashed line ) is uniform ; while in the intermediate region , they spread farther apart as they are trying to keep up with the steeply declining baseline . in the logarithmic scale of fig . 2b , the same peaks appear to come closer together in the two areas where the baseline is flat and become equidistant in the middle area along the gradient of the baseline .
a far more interesting application is shown in figure 3 . the lane
emden equation ( [ main5 ] ) , subject to the physical boundary conditions ( [ bc ] ) , has been integrated numerically for various choices of the four free parameters .
the resulting equilibrium profiles have been optimized for the present solar system assuming that their density maxima correspond to the observed semimajor axes of the planetary orbits out to and including pluto .
not counting the central peak at @xmath69 , the third density maximum is always scaled to a distance of 1 au during this _ nonlinear unconstrained optimization_. in the best
fit model shown in figure [ fig3 ] , the third density peak occurs at @xmath209 , implying that the length scale of the solar nebula in its isothermal phase was quite small ( @xmath210 au ; see also [ solar3 ] below ) .
ccccc + + + index & planet & semimajor & peak & + @xmath13 & name & axis & location & error + & & @xmath14 ( au ) & @xmath211 ( au ) & ( % ) + 1 & mercury & 0.387 & 0.362 & @xmath212 + 2 & venus & 0.723 & 0.705 & @xmath213 + 3 & earth & 1 & 1 & . .
+ 4 & mars & 1.524 & 1.588 & 4.2 + 5 & ceres & 2.765 & 2.686 & @xmath214 + 6 & jupiter & 5.203 & 4.930 & @xmath215 + 7 & saturn & 9.537 & 9.843 & 3.2 + 8 & uranus & 19.19 & 20.09 & 4.7 + 9 & neptune & 30.07 & 29.66 & @xmath216 + 10 & pluto & 39.48 & 39.19 & @xmath217 + 11 & & @xmath218 & 48.73 & + 12 & & @xmath219 & 58.27 & + 13 & eris & 67.89 @xmath220 & 67.80 & @xmath19 + + + + + the density maxima @xmath211 of the best
fit model have been converted to au and are listed in table 2 along with the observed orbital semimajor axes @xmath14 taken from table 1 above .
the new dwarf planet eris was not used in the optimization , but it is also listed at the bottom of table 2 for comparison purposes . in addition ,
table 3 displays all the extrema in the best fitted oscillatory density profile out to 106 au .
table 3 may be useful to observers who are trying to locate more dwarf planets in the outer solar system and to theorists who intend to build more sophisticated models of planitesimal accumulation in the solar nebula .
ccccc + + + index & location & minimum & location & maximum + @xmath13 & of minimum & density & of maximum & density + & @xmath221 ( au ) & @xmath222 & @xmath223 ( au ) & @xmath224 + 1 & 0.17901 & @xmath225 & 0.36162 & @xmath226 + 2 & 0.53124 & @xmath227 & 0.70474 & @xmath228 + 3 & 0.90231 & @xmath229 & 1 & @xmath230 + 4 & 1.4282 & @xmath231 & 1.5875 & @xmath232 + 5 & 2.4182 & @xmath233 & 2.6858 & @xmath234 + 6 & 4.4935 & @xmath235 & 4.9295 & @xmath236 + 7 & 9.7814 & @xmath237 & 9.8431 & @xmath237 + 8 & 15.268 & @xmath238 & 20.093 & @xmath239 + 9 & 24.855 & @xmath240 & 29.661 & @xmath241 + 10 & 34.421 & @xmath242 & 39.193 & @xmath243 + 11 & 43.962 & @xmath244 & 48.726 & @xmath245 + 12 & 53.490 & @xmath246 & 58.266 & @xmath247 + 13 & 63.028 & @xmath248 & 67.800 & @xmath249 + 14 & 72.557 & @xmath250 & 77.336 & @xmath251 + 15 & 82.092 & @xmath252 & 86.857 & @xmath253 + 16 & 91.626 & @xmath254 & 96.399 & @xmath255 + 17 & 101.15 & @xmath256 & 105.91 & @xmath257 + looking at orbital distances interior to the orbit of eris in table 2 , we see that there are two peaks in the model , @xmath258 au and @xmath259 au , in which large dwarf planets have not been discovered .
these two orbital distances lie beyond the outer `` edge '' or `` gap '' of the classical kuiper belt ( 4748 au ; e.g. , delsanti & jewitt 2006 ) , in an area where the number of orbiting objects decreases dramatically ( e.g. , morbidelli , brown , & levison 2003 ) . despite that
, both peaks are located to within @xmath2606% from two large objects : @xmath261 is near makemake ( 2005 fy@xmath262 ) , the third largest classical kuiper belt object after pluto and haumea ; and @xmath263 is near 2002 tc@xmath264 , the second largest scattered disk object after eris .
the relative errors for each individual density peak of the model are also listed in table 2 .
we see that the largest relative error is @xmath212% for the first peak that corresponds to the orbit of mercury .
this deviation of @xmath2600.025 au is still quite small by solar system standards .
the optimization algorithm has minimized the mean relative error @xmath265 for the first 10 orbits listed in table 2 .
this was defined as a `` standard deviation '' by using the square deviations of 9 density peaks and excludes the peak that corresponds to the earth s orbit which has zero deviation because of our scaling assumption that @xmath266 au .
thus : @xmath267 where @xmath268 and the @xmath269 term does not contribute to the sum ( @xmath270 ) . by construction
, this definition of the mean relative error places more weight to the orbital distances of planets near the sun and allows for larger errors in the locations of the outer density peaks . despite this skewing of the fit , the mean relative error for the best
fit model is @xmath271 4.1% , affirming that our simple equilibrium model succeeds in matching all of the observed planetary distances very well .
the parameters of the best
fit model determined by the optimizing algorithm along with the scaling assumption that @xmath272 au ( see table 2 ) constitute a set of important dynamical parameters of the long gone solar nebula : @xmath273
the value of the rotation parameter @xmath274 indicates that the rotation of the isothermal nebula was moderate , about 40% of the maximum value allowed by self gravity .
this result also indicates that the best
fit model ( and the other models discussed here ) is stable to nonaxisymmetric disturbances because @xmath66 is much below the critical value of @xmath275 .
this critical value can be obtained from the @xmath276-parameter criterion for stability of rotating , self gravitating , gaseous systems ( @xmath277 ; christodoulou , shlosman , & tohline 1995 ) by combining the definition @xmath278 with the definition of @xmath276 for disks where @xmath279 to get the relation @xmath280 which implies that @xmath275 .
of course the models discussed in this section are stable to axisymmetric disturbances as well , since they all satisfy the rayleigh criterion . the value of the power law index @xmath281 indicates that the density profile of the differentially rotating region of the nebula declined with radius , on average , as @xmath1 .
this region extended from a radius of @xmath282 au out to a radius of @xmath283 au .
of these parameters , only @xmath183 is slightly uncertain because the mean relative error @xmath265 ( eq.[[error ] ] ) places less weight to the orbits of the outer , nearly equidistant planets which , in effect , determine the truncation radius . as a result
, the optimization procedure also finds additional `` near minima '' of high quality ( @xmath284%4.7% ) , among which the most extreme model has the following parameter values : @xmath285 , @xmath286 , @xmath287 au , @xmath288 ( @xmath289 au ) , and @xmath290 ( @xmath291 au ) . a comparison between these values and the best - fit values listed in eq .
( [ neb ] ) gives us an idea about how shallow the region around the true minimum is in the four parameter space of the model .
note , in particular , that the power law index does not differ from @xmath292 by more than 1.5% and the rotation parameter does not differ from @xmath293 by more than 5% in any of the high quality fits to the data . the cylindrical lane emden equation that we have solved can serve as a model of differentially rotating disks supported by thermal pressure in the @xmath35direction , so we expect that the scale height from pressure support will be @xmath294 down the radial density gradient .
, where @xmath295 is the rotation speed . in our models
, @xmath295 is asymptotically flat when the density exhibits a power law profile and @xmath53 is constant , leading thus to the approximate relation @xmath294 . on the other hand , the scale height is @xmath296 in the core of our best
fit model . as a result ,
the mass estimates calculated in [ md ] for a disk with @xmath297 are too conservative . ] in this case , the value of @xmath292 implies that the corresponding power
law index in the surface density profile ( @xmath298 ) of the nebular disk was @xmath299 .
this value is virtually identical to that obtained by weidenschilling ( 1977 ) who derived an estimate of the surface density distribution of the protoplanetary disk by smearing out the observed planetary masses over annular rings in the disk s midplane and then applied a correction to this distribution by adding the appropriate amount of volatiles in order to bring the elemental abundance of the gas up to solar composition .
our analytical work and modeling effort also provide a direct approach to the same problem , but from a different angle than that conceived by weidenschilling .
the unambiguous congruence of the results obtained by these two disparate methods is rather astonishing and suggests strongly that the surface density profile @xmath300 of the solar nebula was indeed exhibiting an @xmath2 power law in the isothermal phase of its evolution .
furthermore , our work also helps in delineating the fundamental physics behind such a mean surface density profile in the midplane of the solar nebula . with the aid of our best - fit model
, we can deduce substantial new information concerning the structure and the dynamics of the nebular disk .
in addition to the structural and rotation parameters discussed above , we can use our analytic baseline model in order to probe the dynamical state of the protoplanetary disk in its isothermal phase as follows . using the length scale of the disk
( @xmath301 au ) in eq . ( [ length ] ) ,
we can write an equation of state of the form @xmath302 where @xmath53 and @xmath61 may be thought of as the local sound speed and the local density in the inner disk , respectively . for an isothermal gas at temperature @xmath303 , @xmath304 , where @xmath305 is the mean molecular weight and @xmath306 is the universal gas constant .
hence , eq . ( [ crho ] ) can be written in the form @xmath307 where @xmath303 and @xmath305 are measured in degrees kelvin and @xmath308 , respectively . for a cold disk of gas with @xmath309
k and @xmath310 ( molecular hydrogen and neutral helium with fractional abundances @xmath311 and @xmath312 by mass ) , we find that @xmath313 this value is comfortably larger than the well - known threshold for planet formation in the solar nebula ( @xmath314 ; see e.g. , lissauer [ 1993 ] ) and implies that the conditions for planet formation were already in place , at least in the inner disk , in the isothermal phase of the disk s evolution . using the characteristic density of the inner disk ( eq . [ [ therho ] ] ) in the definition of @xmath315
, we can determine the jeans frequency of the disk : @xmath316 then , using the value @xmath317 ( eq.[[neb ] ] ) in the definition of @xmath62 , we can determine the angular velocity of the uniformly rotating core ( @xmath318 au ) : @xmath319 for reference , this value of @xmath42 corresponds to an orbital period of 12 yr . in the _ present solar system _ , that would correspond to a keplerian orbit with semimajor axis @xmath320 au . so the core of the nebula was rotating about as slowly as jupiter is currently revolving around the sun . finally , the constant asymptotic value of the angular velocity @xmath321 for @xmath322 is also a characteristic rotational parameter of the nebula ( because the outer region of uniform density is necessary for the formation of the nearly equidistant outer planets ) .
using eqs .
( [ rot ] ) and ( [ fasym ] ) , we find that @xmath323 where the values listed in eq .
( [ neb ] ) and eq . ( [ theom ] ) were used in the numerical evaluation .
this value of @xmath321 is 28 times smaller than @xmath42 ; and corresponds to an orbital period of 338 yr and to a keplerian orbit with @xmath324 au near the outer edge of the classical kuiper belt in the present solar system .
assuming that our model disk is uniform in the @xmath35direction over a thin layer of thickness @xmath325 , we can multiply its mass per unit length by this factor and we can define the total mass near the midplane of the solar nebula out to a maximum radius @xmath326 as @xmath327 where the constant @xmath328 is given by @xmath329 where eq .
( [ therho ] ) and @xmath210 au were used in the numerical evaluation .
@xmath328 is approximately the mass contained to within one length scale from the center of the disk and implies that the central surface density is @xmath330 using this value , the core radius @xmath331 au , and the power
law index @xmath332 , we can write the surface density profile of the solar nebula for @xmath333 in the form @xmath334 where @xmath36 is measured in au . the value at 1 au is lower than that estimated by weidenschilling ( 1977 ) ; but it agrees very well with hayashi s ( 1981 ) competing result that was obtained by the same method and with the same data , and also with kuchner s ( 2004 ) average estimate obtained from a similar analysis of 26 multiple planet extrasolar systems .
adopting now the baseline density profile ( eq . [ [ den3 ] ] ) as an approximation to the actual density distribution and using the parameters listed in eq .
( [ neb ] ) , we can evaluate the integral of eq .
( [ mass1 ] ) over the three regions of the baseline out to , e.g. , @xmath335 au ( @xmath336 ) : @xmath337 \nonumber \\ & & \nonumber \\ & = & \ \ \ \ \ \ \ \ \ \ \ \ \ \ 1\times 10^{-4 } \
m_\odot \ \ + \ \ 3\times 10^{-4 } \
m_\odot \ \ + \ \ 5\times 10^{-4 } \ m_\odot \\
% & & \nonumber \\ & \approx & \ \ \ \ \ \ \ \ \ \ \ \ \ \ 0.001 \ m_\odot \nonumber \ .\end{aligned}\ ] ] the total mass is one order of magnitude smaller than the low end of the estimate for the `` minimum mass solar nebula '' ( @xmath338 ; weidenschilling 1977 ) .
this is not surprising since we have adopted a very small value ( @xmath40 ) for the vertical scale of the disk . in the classical scenario of cloud collapse ,
this mass near the midplane will be enhanced substantially as more matter from the parent cloud will continue falling onto the disk ( see [ ipc ] below for more details ) .
it is however interesting that the inner disk already has high enough densities ( eq . [ [ therho ] ] ) to begin the process of planet formation so early in its evolution , before the central protostar grows to become gravitationally dominant or radial accretion becomes important in the gas . integrating over the mass distribution of the core of the baseline model , we can estimate important dynamical properties , such as the core mass and angular momentum .
the analytic estimates can then be compared to the corresponding results from protostellar collapse simulations .
such simulations have been recently reviewed by tohline ( 2002 ) who also summarized the main results over which there seems to be widespread agreement among researchers working in the field for almost 40 years ( see the discussion of tohline centered around his figure 2 and table 2 ) .
the mass of the core found in
[ md ] above , @xmath339 , is 40 times smaller than the typical mass of a collapsing cloud core in which the jeans instability may be temporarily halted by thermal pressure and an adiabatic quasi
equilibrium may then be established ( see case b in table 2 of tohline [ 2002 ] ) .
this means that the core of our quasi equilibrium disk will remain isothermal and it will continue to accumulate mass from the infalling cloud beyond the point described here .
furthermore , the total angular momentum of the core of the model is @xmath340 this implies that the _ specific _ angular momentum of the matter in the core is @xmath341 which is about one half of the corresponding estimate given by tohline ( 2002 ) for a cloud core at the endpoint of its isothermal evolution . clearly , in a realistic setting , this type of low
mass core has the potential to grow by accreting a lot more matter of low specific angular momentum , the kind that can settle on to the central region from the vertical direction .
we have presented exact analytic and numerical solutions of the axisymmetric , cylindrically symmetric , isothermal lane emden equation with and without rotation ( eqs . [ [ main3 ] ] and [ [ main2 ] ] , respectively ) .
this second order ode describes the radial equilibrium configurations that are available to a self gravitating perfect fluid in which the _ vertical variation _ of the enthalpy gradient , namely @xmath342 is negligible .
the enthalpy gradient @xmath343 , where @xmath52 is the density and @xmath51 is the pressure , is the term that establishes vertical hydrostatic equilibrium when it successfully balances the gravitational force @xmath344 in the @xmath35direction , but it is its vertical variation that is actually ignored when the cylindrical lane
emden equation is considered .
this term is zero identically for a cylindrical model of infinite vertical extent ; such a model may be applicable to elongated , filamentary , star forming regions in which @xmath345 .
in other astrophysical applications , especially those dealing with gaseous disks that concern us in this work , this second derivative is not identically zero everywhere ; but it is ignored on the basis that it vanishes on the symmetry plane and hopefully it remains negligible away from that plane over some layer of the astrophysical disk ( as in the exact models of schmitz & ebert [ 1986 ] ) . in the specific case of the solar nebula ,
this assumption is probably valid because the disk is not in vertical hydrostatic equilibrium and vertical pressure support is smaller than the weight of the infalling matter everywhere , except near the midplane of the disk where the gradients tend to zero by symmetry anyway .
the rotating analytic solutions are exact intrinsic solutions of the lane
emden equation ( [ main3 ] ) .
their density profiles are all pure power laws or combinations of power laws ( eqs .
[ [ class2 ] ] and [ [ den3 ] ] ) .
these solutions are determined solely by the balance between the _ gradients _ of the centrifugal and the gravitational accelerations ( eq . [ [ gradients ] ] ) because for an isothermal fluid with a power law density profile , the _ radial variation _ of the enthalpy gradient , namely @xmath346 is zero identically ( see eq . [
[ part2 ] ] and the discussion at the end of [ withrot ] ) .
rotation profiles that can support such power law densities are quite general and varied ( eqs .
[ [ class3 ] ] and [ [ class4 ] ] ) .
they contain three integration constants that can be used to compose models of equilibrium disks in which the density profiles are nonsingular and arbitrarily steep in radius .
we have created one such composite equilibrium model and we have applied it to the structure of the midplane of the solar nebula ( [ solar ] ) . in this model , we have connected a central core region of constant density and an outer region of constant density with a power law profile .
the rotation parameter and the size of the core as well as the power law index and the truncation radius of the intermediate region are the four model parameters to be determined based on the presently observed solar system . beyond the truncation radius ,
the density remains constant at a low level .
this turns out to be the necessary nebular background for the formation of nearly equidistant outer planets and its physical origin deserves further investigation .
the density profile of the composite equilibrium disk model is no longer an exact solution when physical boundary conditions are imposed at the center of the disk .
the exact solution to the associated boundary
value problem is obtained by numerical integration that enforces the proper central boundary conditions in a model with the same differential
rotation law as the composite analytic model .
we have found that the intrinsic analytic density profile is a good approximation to the actual numerical solution , i.e. , it is a baseline " that exhibits on average all the important features of the exact solution to the boundary value problem , except one : the actual density profile is permanently oscillatory in radius . in fact , it oscillates around the baseline solution because it is attracted to this inherent particular solution of the lane emden equation , but the two solutions can not match in any finite segment since such coincidence is strictly prohibited by the imposed central boundary conditions ( see fig . 2 and the discussion in [ compo ] ) .
the radial density peaks of the actual solution to the boundary value problem correspond to local minima of the gravitational potential in the midplane of the protoplanetary disk . in a cloud collapse scenario , the density of the gas and the concentration of condensates ( dust , rocks , and ices )
will be enhanced inside these local gravitational potential wells as more matter rains down onto the disk from the surrounding protostellar cloud .
it is not unreasonable to consider that the added material will , in turn , deepen further these potential wells which may thus become feasible sites of planet formation .
based on this picture , we have proceeded to fit the density peaks of the exact numerical solutions of the lane emden boundary value problem to the presently observed locations of the planets around the sun and the satellites of the jovian planets . our best
fit model for the planets in our solar system ( fig . 3 and tables 2 and 3 ) is described in [ solar2 ] and the dynamical parameters obtained for the solar nebula
are described in [ solar3 ] . the mean relative error for this
fit model is 4.1% and the relative deviations for individual planets do not exceed 6.5% .
furthermore , this model suggests that the radial surface density profile of the protoplanetary disk between roughly 0.8 au and 11.3 au declined as @xmath2 , a result that agrees fully with the classic determinations of weidenschilling ( 1977 ) and hayashi ( 1981 ) which were based on the presently observed planetary masses .
we find the confluence of these very diverse approaches encouraging and supportive of the view that the fundamental notions behind our model are indeed relevant to the problem of the radial distribution of planetary orbits .
we would also like to note that the same type of modeling works very well for the regular satellites of jupiter and for the 5 planets of 55 cancri , and we plan to present these results in future publications .
our work is the first to produce exact solutions of the nonlinear isothermal lane
emden equation with differential rotation that exhibit a pronounced oscillatory behavior , a feature attributed exclusively to the proper choice of the central boundary conditions .
the roughly arithmetic or geometric spacings of the corresponding density maxima depend on the local slope of the differential rotation profile and provide a transparent , physically based , and soundly formulated reason for producing a sequence of distinct density enhancements at radial positions that agree with the present planetary distances , given that the latter are observed to follow an arithmetic , then a geometric , and finally again an arithmetic progression ( see table 1 ) .
hence , it should be remarked that we now have at hand , for the first time , a plain explanation of the so - called titius bode
law " of planetary distances .
this empirical algorithm has been known for 240 years but with no underlying physical justification ( see e.g. graner & dubrulle [ 1994 ] and hayes & tremaine [ 1998 ] ) .
fit model of the midplane of the solar nebula ( fig . 3 and table 2 ) shows a progression of density enhancements that are stretched farther apart from one another , as the actual density profile decreases sharply in trying to keep up with the steeply declining baseline .
the relative spacing of the peaks is roughly geometric between 0.7 and 20 au , and this explains the success of the titius bode algorithm for a large number of planetary orbits .
the change of the underlying density profile to constant beyond a certain distance in our model can then also explain why this algorithm fails at large distances .
a similar account of the nongeometric spacing of mercury also finds a straightforward explanation in terms of a uniform central core in our model .
so , just as many researchers have suspected in the past ( see [ tb ] ) , the titius bode algorithm has no underlying physical principle underneath its phenomenology ; the algorithm has been successful only to the extent that it has managed to exploit empirically the most pronounced feature of the radial density profile of the solar nebula ( the roughly geometric spacing of the peaks between 0.7 and 20 au ) . finally , we should point out that the above resolution of the long standing problem of planetary `` order '' in our solar system is very anticlimactic , as the long sought physical explanation turns out to lie entirely within the realm of conventional physics and does not need to invoke unrestrained numerology of the titius bode type , exotic physics , `` new '' dynamical laws , or arbitrary `` universal '' constants and ad hoc solar system `` quantizations '' such as those proposed in the past .
we view this work only as a first ( but compelling ) step toward understanding the systematic formation of planets ; the evolution that results in the formation of the entire solar system likely involves additional processes and physics ( e.g. , dissipative disk accretion , magnetic fields ) that are necessary in delineating its present structure and composition .
however , we believe that the condensed solid cores inside the gravitational potential minima of the solar nebula could possibly survive the subsequent evolution and could form planets at the same orbital locations . furthermore , we hope that the results presented here will provide the motivation behind the formulation of some novel hypotheses concerning the formation and the dynamical evolution of certain sectors in the early solar system and in the recently discovered extrasolar planetary systems ( see the reviews of lissauer [ 1993 ] , beckwith & sargent [ 1996 ] , and marcy et al . [ 2005 ] ) . some of the ideas that emerge from our study are outlined below .
our results support the segregation of condensates at specific radii in which the gas density is larger than the mean background value ( fig . 3 and table 2 ) .
these orbital locations in the midplane of the disk are ideal sites for growing planitesimals by accumulation of smaller bodies in a systematic ( nonrandom ) way . in the best
fit density profile @xmath347 with @xmath292 , the absolute spacing of these sites of concentrated material depends on a single parameter , the length scale of the disk ( @xmath210 au ) which is a measure of the entropy content of the gas ( see eq . [ [ crho ] ] ) .
the segregation of condensates in initially shallow ( see table 3 ) gravitational potential wells should be tested by computational experiments because this kind of work may potentially lead to improved protoplanetary models .
numerical work is quite common in investigations of the early solar system but the simulations have always started from arbitrary initial conditions ( e.g. , wetherill 1989 ; ruzmaikina , safronov , & weidenschilling 1989 ; bodenheimer , ruzmaikina , & mathieu 1993 ; boss 1995 ) because of the limited information that can be collected from observations , meteorites , and laboratory experiments .
our results can potentially provide a better handle on the appropriate initial conditions for simulations of this type .
numerical investigations of planet migration have intensified over the past 10 years ( see the progress report by levison et al . [ 2007 ] ) in an attempt to understand the complex dynamics observed in the kuiper belt : the trapping of some small objects in the mean motion resonances of neptune , the gravitational scattering of other small objects to high inclinations , the severe mass deficiency ( by at least a factor of 100 ) of the entire kuiper belt , and the abrupt outer edge " or gap " of the classical kuiper belt at 4748 au ( e.g. , delsanti & jewitt 2006 ) .
our results do not support the idea that the gaseous giant planets have secularly migrated over time to the presently observed radial locations ( fernndez & ip 1984 ; malhotra 1993 , 1995 ; gomes , morbidelli , & levison 2004 ; gomes et al . 2005 ) . the quality of our best
fit model of planetary distances is so high ( see [ solar2 ] ) that it seems improbable that any planet that scattered small planitesimals has managed to jump out of its local gravitational potential well and has moved to a new location inside an adjacent potential well . on the other hand
, the displacement of a planet within its potential well appears viable in the context of our model , and the potential wells of the gaseous giants are quite large in size ( see table 3 and fig .
3a ) : uranus and neptune are both orbiting on potential minima of radial half
width @xmath348 au ; saturn s minimum has @xmath349 au ; and jupiter s minimum has @xmath350 au .
these values are not very different than the migration distances used so far in the numerical simulations ; only saturn presents a challenge because its potential well is too steep at smaller radii to allow for an outward migration of more than @xmath2600.1 au to its present location .
perhaps new models of planet migration can be constructed in which these results will be taken into account .
fig . 3b and table 3
show that the first density minimum in the best
fit model of the solar nebula occurs at @xmath351 au .
this suggests that there were no likely planetary sites interior to the orbit of mercury .
it also suggests that there was a significantly lower concentration of condensates within this minimum , in the area below the baseline that extends approximately from 0.1 au to 0.3 au . since we expect that our solar system is in no way special but rather representative of large planetary systems around solar type stars , we believe that the same deficiency of solids should also exist in the inner regions of other protostellar disks that are currently in the process of forming their protostars and protoplanets .
we think that such an inner gap in condensates is currently inferred in the circumstellar disks of some pre
main sequence stars ; these disks show almost no near infrared radiation emanating from the inner @xmath352 au ( strom , edwards , & skrutskie 1993 ; beckwith & sargent 1993 ; millan gabet et al .
2001 ; akeson et al . 2005 ; fedele et al . 2008 ) .
the alternative view is that an orbiting protoplanet may have cleared a gap in that area .
although this may be possible in small heavy disks where large planets could form very close to their stars ( christodoulou & kazanas 2008 ) , it can not explain large systems like our own in which gas giants are not expected to form within the inner 0.3 au .
it is believed on theoretical grounds that terrestrial planets are commonly formed near their central stars and that it is unlikely to have a gaseous giant form closer to the center than about 4 au ( boss 1995 ) . accordingly , our solar system is presumed to be typical of planetary systems around other stars . on the other hand , most of the extrasolar planets found to date do not fit in this theoretical picture .
the currently available sample of extrasolar planets is strongly biased toward giant planets because small perturbing masses can not be easily resolved by observations , so it is understood that it will take more time before large extrasolar systems comparable to our own can be detected .
nevertheless , there is presently no doubt that in many extrasolar systems , a large number of massive planets exist very close to their stars ( the trail of discovery started with the planet at 0.05 au in 51 peg found by mayor & queloz [ 1995 ] and the planet at 0.48 au in 70 vir found by marcy & butler [ 1996 ] ) .
this discrepancy seems to point to the hypothesis that the cores of these planets have formed farther away and that the objects have migrated inward to the observationally inferred distances ( kary & lissauer 1995 ; lin , bodenheimer , & richardson 1996 ) , perhaps destroying in the process the orbits of terrestrial planets that could have existed closer to central stars .
migrating giant planets can not be accounted for in our model , so we will have to wait until observations can detect some large extrasolar systems similar to our own . however , two small multiple
planet systems have already been detected which are different than our solar system , but their planetary distributions appear to be more in line with a sequence of well ordered density peaks as predicted by our model .
these systems are 55 cancri ( fischer et al .
2008 ) and hd 37124 ( vogt et al .
2005 ) . using updated data from the `` catalog of nearby exoplanets '' ( http://exoplanets.org and butler et al . [ 2006 ] ) ,
we find that the ratios of semimajor axes of the three innermost planets in these two systems are @xmath353 and @xmath354 .
these ratios are barely larger than those in our solar system ( 1.9 and 1.4 , respectively ) .
in contrast , all the other systems in the catalog with three or more exoplanets have @xmath355 and/or @xmath356 , ratios that are too large compared to those in 55 cancri , hd 37124 , and our solar system .
we see then that the relative distributions of planetary orbits in 55 cancri and hd 37124 are effectively the same , despite the fact that these two well ordered systems are much smaller than our solar system and their central stars have widely different chemical compositions ( 55 cancri is very metal rich , while hd 37124 is very metal poor ) .
55 cancri is now believed to have 5 planets in remarkably circular orbits ( fischer et al .
2008 ) , and this number is enough to allow for our type of modeling for this system .
we have carried out the analysis of 55 cancri and we present our results in a companion paper ( christodoulou & kazanas 2008 ) .
akeson , r. l. , boden , a. f. , monnier , j. d. , et al .
2005 , , 635 , 1173 beckwith , s. v. w. , & sargent , a. i. 1993 , in protostars and planets iii , ed . e. h. levy , & j. i. lunine ( tucson : univ . of arizona press ) , 521 beckwith , s. v. w. , & sargent , a. i. 1996 , nature , 383 , 139 bender , c. m. , & orzag , s. a. 1978 , advanced mathematical methods for scientists and engineers , ( new york : mcgraw hill ) berkovich , l. m. 1997 , symm . nonlinear math phys , 1 , 155 binney , j. , & tremaine , s. 1987 , galactic dynamics ( princeton : princeton univ . press ) bodenheimer , p. , ruzmaikina , t. , & mathieu , r. s. 1993 , in protostars and planets iii , ed . e. h. levy , & j. i. lunine ( tucson : univ . of arizona press ) , 367 boss , a. p. 1995 , science , 267 , 360 brown , m. e. , trujillo , c. a. , & rabinowitz , d. l. 2005 , , 635 , l97 butler , r. p. , & marcy , g. w. 1996 , , 464 , l153 butler , r. p. , wright , j. t. , & marcy , g. w. , et al .
2006 , , 646 , 505 chandrasekhar , s. 1939 , an introduction to the study of stellar structure , ( chicago : univ . of chicago press ) christodoulou , d. m. 1993 , , 412 , 696 christodoulou , d. m. , contopoulos , j. , & kazanas , d. 1996 , , 462 , 865 christodoulou , d. m. , & kazanas , d. 2008 , , submitted christodoulou , d. m. , & narayan , r. 1992 , , 388 , 451 christodoulou , d. m. , shlosman , i. , & tohline , j. e. 1995 , , 443 , 551 coxeter , h. s. m. 1989 , introduction to geometry , 2nd edition ( new york : wiley ) danby , j. m. a. 1988 , fundamentals of celestial mechanics , 2nd edition ( richmond : willmann bell ) delsanti , a. , & jewitt , d. 2006 , in solar system update , ed .
blondel & j. w. mason ( berlin : springer praxis ) , 267 dixon , j. m. , & tuszyski , j. a. 1990 , phys .
a , 41 , 4166 dubrulle , b. , & graner , f. 1994 , a&a , 282 , 269 emden , r. 1907 , gaskugeln ( leipzig : b. g. teubner ) fabbiano , g. 1989 , , 27 , 87 fabian , a. c. 1991 , a&a rev . , 2 , 191 fedele , d. , van den ancker , m. e. , acke , b. , et al .
2008 , a&a , preprint ( astro - ph/0809.3947 ) fermi , e. 1927 , rend .
lincei , 6 , 602 fernndez , j. a. , & ip , w.h .
1984 , icarus , 58 , 109 fischer , d. a. , marcy , g. w. , butler , r. p. , et al .
2008 , , 675 , 790 fowler , r. h. 1914 , quart . j. math .
, 45 , 289 fowler , r. h. 1930 , , 91 , 63 fowler , r. h. 1931 , quart . j. math . , 2 , 259 frank
kamenetskii , d. a. 1955 , diffusion and heat exchange in chemical kinetics ( princeton : princeton univ .
press ) goenner , h. 2001 , gen .
rel , grav . , 33 , 833 goenner , h. , & havas , p. 2000 , j. math .
phys . , 41 , 7029 gomes , r. s. , morbidelli , a. , & levison , h. f. 2004 , icarus , 170 , 492 gomes , r. s. , gallardo , t. , fernndez , j. a. , & brunini , a. 2005 , celest .
astron . , 91 , 109 goodman , j. , & narayan , r. 1988 , , 231 , 97 graner , f. , & dubrulle , b. 1994 , a&a , 282 , 262 hansen , c. j. , aizenman , m. l. , & ross , r. r. 1976 , , 207 , 736 hayashi , c. 1981 , prog .
, 70 , 35 hayes , w. , & tremaine , s. 1998 , icarus , 135 , 549 horedt , g. p. 1986
, a&a , 160 , 148 jaki , s. 1972 , am .
j. phys . , 40 , 93 jeans , j. h. 1914 , phil .
royal soc .
london , series a , 213 , 457 kary , d. m. , & lissauer , j. j. 1995 , icarus , 117 , 1 kaufmann , w. j. , iii 1994 , universe , 4th edition ( new york : freeman ) kochanek , c. s. 1995 , , 445 , 559 kuchner , m. j. 2004 , , 612 , 1147 lane , l. j. h. 1869 - 70 , amer . j. sci .
arts , 4 , 57 lecar , m. 1973 , nature , 242 , 318 levison , h. f , morbidelli , a. , gomes , r. s. , & backman , d. 2007 , in protostars and planets v , ed .
b. reipurth , d. jewitt , & k. keil ( tucson : univ . of arizona press ) , 669 li , x. q. , zhang , h. , & li , q. b. 1995 , a&a , 304 , 617 lin , d. n. c. , bodenheimer , p. , & richardson , d. c. 1996 , nature , 380 , 606 lions , p. l. 1982 , siam rev . , 24 , 441 lissauer , j. j. 1993 , , 31 , 129 lynch , p. 2003
, , 341 , 1174 malhotra , r. 1993 , nature , 365 , 819 malhotra , r. 1995 , , 110 , 420 marcy , g. w. , & butler , r. p. 1996
, , 464 , l147 marcy , g. w. , butler , r. p. , fischer , d. a. , et al . 2005 , prog .
suppl . , 158 , 24 mayor , m. , & queloz , d. 1995 , nature , 378 , 355 millan gabet , r. , schloerb , f. p. , & traub , w. a. 2001 , , 546 , 358 morbidelli , a. , brown , m. e. , & levison , h. f. 2003 , earth , moon & planets , 92 , 1 narita , s. , kiguchi , m. , miyama , s. m. , & hayashi , c. 1990 , , 244 , 349 neuh@xmath10user , r. , & feitzinger , j. 1986 , a&a , 170 , 174 nieto , m. m. 1972 , the titius bode law of planetary distances : its history and theory ( oxford : pergamon press ) nottale , l. , schumacher , g. , & gay , j. 1997 , a&a , 322 , 1018 ostriker , j. 1964 , , 140 , 1056 rix , h - w , de zeeuw , p. t. , cretton , n. , van der marel , r. p. , & carollo , c. m. 1997 , , 488 , 702 robe , h. 1968 , ann .
dastrophys . , 31 , 549 ruzmaikina , t. v. , safronov , v. s. , & weidenschilling , s. j. 1989 , in asteroids ii , ed . r. p. binzel , t. gehrels , & m. s. matthews ( tucson : univ . of arizona press ) , 681 sarazin , c. l. 1988 , x ray emission from clusters of galaxies ( cambridge : cambridge univ .
press ) schmitz , f. 1984 , a&a , 131 , 309 schmitz , f. , & ebert , r. 1986 , a&a , 154 , 214 schneider , m. , & schmitz , f. 1995 , a&a , 301 , 933 sornette , d. 1998 , phys . rep . , 297 , 239 stodlkiewicz , j. s. 1963 , acta astron .
, 13 , 30 strom , s. e. , edwards , s. , & skrutskie , m. f. 1993 , in protostars and planets iii , ed . e. h. levy , & j. i. lunine ( tucson : univ . of arizona press ) , 837 thomas , l. h. 1927 , proc .
cambridge phil .
soc . , 23 , 542 tohline , j. e. 2002 , , 40 , 349 visser , m. , & yunes , n. 2003 , int . j. mod . phys . , 18 , 1 vogt , s. s. , butler , r. p. , marcy , g. w. , et al .
2005 , , 632 , 638 weidenschilling , s. j. 1977 , a&ss , 51 , 153 wetherill , g. w. 1989 , in asteroids ii , ed .
r. p. binzel , t. gehrels , & m. s. matthews ( tucson : univ . of arizona press ) , 661 wong , j. s. w. 1975 , siam rev . , 17 , 339 | we have derived exact solutions of the isothermal lane emden equation with and without rotation in a cylindrical geometry .
the corresponding hydrostatic equilibria are relevant to the dynamics of the solar nebula before and during the stages of planet and satellite formation .
the nonrotating solution for the mass density is analytic , nonsingular , monotonically decreasing with radius , and it satisfies easily the usual physical boundary conditions at the center . when differential rotation is added to the lane emden equation , a new class of exact solutions for the mass density appears . we have determined all of these solutions analytically as well . within this class , solutions that are
power laws or combinations of power laws are not capable of satisfying the associated boundary value problem , but they are nonetheless of profound importance because they constitute `` baselines '' to which the actual solutions approach when the central boundary conditions are imposed . numerical integrations that enforce such physical boundary conditions show that the actual radial equilibrium density profiles emerge from the center close to the nonrotating solution , but once they cross below the corresponding baselines , they cease to be monotonic . the actual solutions are forced to oscillate permanently about the baseline solutions without ever settling onto them because the central boundary conditions strictly prohibit the matching of the two types of solutions .
this oscillatory behavior of the isothermal solutions to the lane emden boundary value problem is entirely generic and extends to polytropic models as well .
based on our results , we expect that quasistatically evolving protoplanetary disks should develop oscillatory radial density profiles in their midplanes during the isothermal phase of their collapse .
the peaks in these profiles correspond to local gravitational potential minima and their radial locations are ideal sites for the formation of protoplanets ; sites that can be accentuated during infall of more matter from the still collapsing cloud .
indeed , a straightforward application to our solar system using an oscillatory solution derived from a differentially rotating baseline yields a highly accurate match ( mean relative error @xmath0 ) between the radial density peaks of the model and the semimajor axes of all the major planets and the dwarf planets , provided that the mean density profile between 0.8 au and 11.3 au falls off with radius as @xmath1 ( corresponding to a mean surface density variation of @xmath2 that is consistent with the profile determined empirically by weidenschilling in 1977 ) .
we believe then that for the first time in over two centuries we have a mathematically rigorous explanation of all planetary orbits in our solar system and the physics that is responsible for planet formation at radii that are not in the least random or arbitrary . |
the normalized colored jones polynomial @xmath0 for a link @xmath1 is a sequence of laurent polynomials in the variable @xmath2 , i.e. @xmath3 $ ] .
this sequence is defined for @xmath4 so that @xmath5 is the ordinary jones polynomial , and @xmath6 where @xmath7 is the unknot . for links @xmath1 with an odd number of components ( including all knots ) , @xmath8 is actually in @xmath9 $ ] . for links with an even number of components , @xmath10 $ ] . in @xcite and
@xcite oliver dasbach and xiao - song lin showed that , up to sign , the first two coefficients and the last two coefficients of @xmath11 do not depend on @xmath12 for alternating knots .
they also showed that the third ( and third to last ) coefficient does not depend on @xmath12 so long as @xmath13 .
this and computational data led them to believe that the @xmath14-th coefficient does not depend on @xmath12 so long as @xmath15 .
the goal of this paper is to prove this conjecture for all alternating links .
it is also known that this property of @xmath0 does not hold for all knots . in @xcite , with oliver dasbach , we examined the case of the @xmath16 torus knot for which this property fails . for two laurent series @xmath17 and @xmath18
we define @xmath19 if after multiplying @xmath17 by @xmath20 and @xmath18 by @xmath21 , @xmath22 and @xmath23 some powers , to get power series @xmath24 and @xmath25 each with positive constant term , @xmath24 and @xmath25 agree @xmath26 . for example @xmath27
another way of phrasing the above definition is that @xmath28 if and only if their first @xmath29 coefficients agree up to sign . in @xcite we defined two power series , the head and tail of the colored jones polynomial @xmath30 and @xmath31 .
the tail of the colored jones polynomial of a link @xmath1 if it exists is a series @xmath31 , with @xmath32 similarly , the head @xmath30 of the colored jones polynomial of @xmath1 is the tail of @xmath33 , which is equal to the colored jones polynomial of the mirror image of @xmath1 .
note that @xmath31 exists if and only if @xmath34 for all @xmath12 .
for example , for the first few colors @xmath12 the colored jones polynomial of the knot @xmath35 multiplied by @xmath36 is ll @xmath37 & @xmath38 + @xmath39 & @xmath40 + @xmath41 & @xmath42 + @xmath43 & @xmath44 + @xmath45 & @xmath46 + @xmath47 & @xmath48 + this is exactly the property conjectured by dasbach and lin to hold for all alternating knots , and the subject of the main theorem of this paper .
[ thmalt ] if @xmath1 is an alternating link , then @xmath34 .
because the mirror image of an alternating link is alternating , this theorem says that the head and the tail exists for all alternating links .
theorem [ thmalt ] was simultaneously and independently proved by stavros garoufalidis and thang le in @xcite using alternate methods .
we are also able to prove a more general theorem about @xmath49-adequate links .
[ thmad ] if @xmath1 is a @xmath49-adequate link , then @xmath34 .
because all alternating links are @xmath49-adequate , theorem [ thmad ] implies theorem [ thmalt ] . in section [ background ]
, we discuss definitions and basic results regarding adequate links and skein theory . in section [
main ] , we present the main lemma which is a slight generalization of a lemma in @xcite relating the lowest terms of the colored jones polynomial to a certain skein theoretic graph .
finally , in section [ proof ] , we present the proofs of theorems [ thmalt ] and [ thmad ] using the graph discussed in section [ main ] .
i would like to thank oliver dasbach for all his help and advise .
i would also like to thank pat gilmer for teaching me all i know about skein theory .
given a link diagram @xmath50 there are two ways to smooth each crossing , described in figure [ smoothings ] . a state of the diagram is a choice of smoothing for each crossing .
two states are particularly important when dealing with the colored jones polynomial ; they are the all - a state @xmath51 and the all - b state @xmath52 .
the all - a ( respectively all - b ) state is the state for which the a ( b ) smoothing is choosen for every crossing . for a state @xmath53
we can build a graph @xmath54 called the state graph for @xmath53 .
the graph @xmath54 has vertices the circles in @xmath53 , and edges the crossings in @xmath50 .
each edge connects the two vertices corresponding to the two circles that the crossing meets .
a link diagram is a - adequate ( b - adequate ) if the state graph for @xmath51 ( @xmath52 ) has no loops .
a link diagram is adequate if it is both a and b - adequate .
a link is adequate if it has an adequate diagram .
the most important property of a - adequate diagrams is that the number of circles in @xmath51 is a local maximum . in other words ,
any state that has only a single @xmath55 smoothing will have one fewer circle than the all - a smoothing .
similarly for b - adequate diagrams , that the number of circles in @xmath52 is a local maximum . it is a well - known fact that all alternating links are adequate links . in particular , a reduced alternating diagram , that is an alternating diagram without any nugatory crossings , is an adequate diagram .
another important fact about adequate diagrams is that parallels of a - adequate diagrams are also a - adequate diagrams . given a diagram @xmath50 , the @xmath56-th parallel of @xmath50 denoted @xmath57 is the diagram formed by replacing @xmath50 with @xmath56 parallel copies of @xmath50 . for a more detailed explanation of skein theory ,
see @xcite or @xcite the kauffman bracket skein module , @xmath58 , of a @xmath59-manifold @xmath60 and ring @xmath61 with invertible element @xmath49 , is the free @xmath61-module generated by isotopy classes of framed links in @xmath60 , modulo the submodule generated by the kauffman relations : [ cols="^,^ " , ] with the properties @xmath62{idemsquared.pdf } \raisebox{.5in}{$\;=\;$ } \includegraphics[height=1in]{idem.pdf}\raisebox{.5 in } { , } \ , \qquad \qquad \includegraphics[height=1in]{idemzero.pdf } \raisebox{.5in}{$\;= 0 $ } \raisebox{.5in}{,}\ , \qquad \qquad \includegraphics[height=1in]{idemloop.pdf } \raisebox{.45in}{$\;= \left ( \frac { \delta_{n+1 } } { \delta_{n } } \right ) \;$ } \includegraphics[height=1in]{idem2.pdf}\ ] ] if @xmath63 , define @xmath64 as the coefficient of @xmath60 in the expansion of the jones - wenzl idempotent . thus @xmath65 .
if @xmath66 is the identity element of @xmath67 , then @xmath68 .
[ idem ] if @xmath63 , then @xmath69 is at least twice the minimum word length of @xmath60 in terms of the @xmath70 s .
this follows easily from the recursive definition of the idempotent by an inductive argument .
the only issue is that terms of the form @xmath71 \draw ( 0,0 ) node[right]{\small{n}}--(0,1 ) node[left]{\scriptsize{n-1 } } --(0,2 ) node[right]{\small{n } } ; \draw[ultra thick , fill = white ] ( -0.2 , 0.5 ) rectangle ( 0.4 , 0.65 ) ; \draw ( 0.2 , 0.65 ) arc ( 180:0:0.25 ) ; \draw ( 0.7,0 ) node[right]{1}--(0.7 , 0.65 ) ; \draw[ultra thick , fill = white ] ( -0.2 , 1.5 ) rectangle ( 0.4 , 1.35 ) ; \draw ( 0.2 , 1.35 ) arc ( -180:0:0.25 ) ; \draw ( 0.7,2 ) node [ right]{\small{1 } } --(0.7 , 1.35 ) ; \end{tikzpicture}$ ] may have a circle which needs to be removed .
in this situation , the minimum degree of the coefficient is reduced by two , but the number of generators used is also reduced by one .
using lemma [ idem ] we can find a lower bound for the minimum degree of any element of @xmath72 which contains the jones - wenzl idempotent . before we do this , consider a crossing - less diagram @xmath53 in the plane consisting of arcs connecting jones - wenzl idempotents .
we will define what it means for such a diagram to be adequate in much the same way that a knot diagram can be a or b- adequate . construct a crossing - less diagram @xmath73 from @xmath53 by replacing each of the jones - wenzl idempotents in @xmath53 by the identity of @xmath74 .
thus @xmath73 is a collection of circles with no crossings .
consider the regions in @xmath73 where the idempotents had previously been .
@xmath53 is adequate if no circle in @xmath73 passes through any one of these regions more than once .
figure [ fig : adequate ] shows an example of a diagram that is adequate and figure [ fig : inadequate ] shows an example of a diagram that is not adequate . in both figures every arc is labelled @xmath75 .
if @xmath53 is adequate , then the number of circles in @xmath73 is a local maximum , in the sense that if the idempotents in @xmath53 are replaced by other elements of @xmath67 such that there is exactly one hook total in all of the replacements , then the number of circles in this diagram is one less than the number of circles in @xmath73 .
if the diagram @xmath53 happens to have crossings in it , we can still construct the diagram @xmath73 , which is now a link diagram . denote @xmath76 .
[ replace ] if @xmath77 is expressed as a single diagram containing the jones - wenzl idempotent , then @xmath78 .
if the diagram for @xmath53 is a crossing - less adequate diagram , then @xmath79 .
first suppose that the diagram @xmath53 has no crossings .
we can get an expansion of @xmath53 by expanding each of the idempotents that appear in the diagram . consider a single term @xmath80 in this expansion . unless all of the idempotents have been replaced by the identity in this term
, then there will be a hook somewhere in the diagram . by removing a single hook
, we get a different term @xmath81 in the expansion .
the number of circles in @xmath80 differs from the number of circles in @xmath81 by exactly one .
also there are fewer hooks in @xmath81 ,
so by lemma [ idem ] and the fact that removing a circle results in multiplying by @xmath82 , the minimum degree of @xmath80 is at least as large as the minimum degree of @xmath81 .
this tells us that the lowest degree amongst terms in the expansion of @xmath53 is the degree of the term with the idempotents replaced by the identity , @xmath73 .
if @xmath53 is adequate , then for any term @xmath80 with only a single hook , @xmath81 will be @xmath73 , and thus @xmath81 will have one more circle than @xmath80 .
therefore , @xmath83 .
this tells us that any term @xmath84 in this expansion will have @xmath83 , and thus @xmath85 .
if there were crossings in @xmath53 , then we can get an expansion of @xmath53 by expanding the idempotents that appear in @xmath53 and summing over all possible smoothings of the crossings .
if we expand over the smoothings first , we get a collection of terms each of which is a coefficient times a crossingless diagram with idempotents .
we can apply the previous argument to say that the minimum degree of each term is the minimum degree of that term with the idempotents replaced with the identity .
now consider @xmath73 . by expanding @xmath73 by summing over all possible smoothings of @xmath73
, we get the same sum as before .
thus the minimum degree of @xmath73 agrees with the minimum degree of @xmath53 .
we can use trivalent graphs to express elements in a skein module using the following correspondence : @xmath62{vertex.pdf } \raisebox{.5in}{$\;=\;$ } \includegraphics[height=1in]{vertexsmooth.pdf}\ ] ] fusion is given by @xmath86 \draw ( 0.7,0 ) .. controls ( 0.2 , 0.6 ) and ( 0.2,1.4 ) .. ( 0.7,2 ) node[right]{\scriptsize{b } } ; \draw ( 0.35,1 ) node[shape = rectangle , draw , ultra thick , fill = white ] { } ; \draw ( -0.7,0 ) .. controls ( -.2,0.6 ) and ( -0.2,1.4 ) .. ( -0.7,2 ) node[left]{\scriptsize{a } } ; \draw ( -0.35,1 ) node[rectangle , draw , ultra thick , fill = white ] { } ; \end{tikzpicture } = \sum_{c } \frac { \delta_c } { \theta(a , b , c ) } \begin{tikzpicture}[baseline=0.8 cm ] \draw ( 1,0 ) node[right]{\scriptsize{b } } --(0.5,0.5 ) ; \draw ( 0,0 ) node[left]{\scriptsize{a } } --(0.5,0.5 ) ; \draw ( 0.5,0.5)--(0.5,1 ) node[right]{c}-- ( 0.5,1.5 ) ; \draw ( 0.5,1.5)--(0,2 ) node[left ] { \scriptsize{a } } ; \draw ( 0.5,1.5)--(1,2 ) node[right ] { \scriptsize{b } } ; \fill ( 0.5,0.5 ) circle ( 0.07 ) ; \fill ( 0.5,1.5 ) circle ( 0.07 ) ; \end{tikzpicture}\ ] ] where the sum is over all @xmath87 such that : 1 .
@xmath88 is even 2 .
@xmath89 to define @xmath90 let @xmath91 and @xmath87 related as above and @xmath92 and @xmath93 be defined by @xmath94 and @xmath95 then @xmath96 \draw ( 0,1 ) ellipse ( 0.6 and 1 ) ; \draw ( 0.7,0.96 ) node[right]{\scriptsize{c } } ; \draw ( -0.7,0.96 ) node[left]{\scriptsize{a } } ; \draw ( 0,0)--(0,1 ) node[right]{\scriptsize{b}}--(0,2 ) ; \fill ( 0,0 ) circle ( 0.07 ) ; \fill ( 0,2 ) circle ( 0.07 ) ; \end{tikzpicture}\ ] ] and one can show that @xmath97 furthermore one has : @xmath98 \draw(0,0)-- ( -.5,.5 ) -- ( -.1,0.9 ) ; \draw ( .1,1.1)--(.5,1.5 ) node[right]{\scriptsize{b } } ; \draw ( 0,0 ) -- ( .5,.5)--(-.5,1.5 ) node[left]{\scriptsize{a } } ; \draw ( 0,0)--(0,-0.8 ) node[right]{\scriptsize{c } } ; \fill ( 0,0 ) circle ( 0.07 ) ; \end{tikzpicture } = ( -1)^{\frac{a+b - c } 2 } a^{a+b - c+\frac{a^2+b^2-c^2}{2 } } \begin{tikzpicture}[baseline=.5 cm , rounded corners=2 mm ] \draw ( 0,0)--(0,-0.8 ) node[right]{\scriptsize{c } } ; \draw ( 0,0)--(0.5,1.5 ) node[right]{\scriptsize{b } } ; \draw ( 0,0)--(-0.5,1.5 ) node[left]{\scriptsize{a } } ; \fill ( 0,0 ) circle ( 0.07 ) ; \end{tikzpicture}\ ] ] we are only interested in the list of coefficients of the colored jones polynomial .
in particular we consider polynomials up to powers of their variable .
up to a factor of @xmath99 for some power @xmath100 that depends on the writhe of the link diagram the ( unreduced ) colored jones polynomial @xmath101 of a link @xmath1 can be defined as the value of the skein relation applied to the link were every component is decorated by an @xmath29 together with the jones - wenzl idempotent .
recall that @xmath102 . to obtain the reduced colored jones polynomial we have to divide @xmath101 by its value on the unknot .
in this section we will relate the tail of the colored jones polynomial to a certain trivalent graph viewed as an element of the kauffman bracket skein module of @xmath104 .
this construction was used in @xcite to prove interesting properties of the head and the tail of the colored jones polynomial . given an b - adequate diagram @xmath50 of a link @xmath1 , consider a negative twist region .
apply the identities of section [ background ] to get the equation : @xmath105 \draw ( -.5,2 ) -- ( -0.1 , 2.4 ) ; \draw ( 0.1,2.6 ) -- ( 0.5,3 ) ; \draw ( 0.5 , 2 ) -- ( -0.5,3 ) ; \draw[dashed ] ( 0 , 1.5 ) -- ( 0,1.75 ) node[left]{\small{m } } --(0 , 2 ) ; \draw ( -.5 , -0.5 ) node[left]{\tiny{n } } -- ( -0.1,-0.1 ) ; \draw ( .5,-0.5 ) node[right]{\tiny{n}}-- ( 0,0)-- ( -.5,.5 ) -- ( -.1,0.9 ) ; \draw ( .1,1.1)--(.5,1.5 ) ; \draw ( 0.1,0.1 ) -- ( .5,.5)--(-.5,1.5 ) ; \end{tikzpicture } \ , = \sum_{j=0}^n ( \gamma(n , n,2j))^m \frac{\delta_{2j}}{\theta(n , n,2j ) } \begin{tikzpicture}[scale=0.6 , baseline=0.8 cm ] \draw ( -.5,3 ) node[left]{\tiny{n}}-- ( 0 , 2.5 ) ; \draw ( 0.5,3 ) node[right]{\tiny{n}}-- ( 0,2.5 ) ; \draw ( 0 , 2.5 ) --(0,1.25 ) node[right]{\tiny{2j}}-- ( 0,0 ) ; \draw ( 0,0)--(-.5,-.5 ) node[left]{\tiny{n } } ; \draw ( 0,0)--(.5,-.5 ) node[right]{\tiny{n } } ; \end{tikzpicture } .\ ] ] here @xmath106 if we apply this equation to every maximal negative twist region , then we get an embedded trivalent graph called @xmath107 .
we get a colored graph @xmath108 where @xmath14 is the number of maximal negative twist regions and @xmath109 by coloring the edges coming from the @xmath110-th twist region by @xmath111 and coloring all of the other edges by @xmath29 . from the previous equation ,
it is clear that we get @xmath112 the following theorem is a useful tool to find properties of the head and tail of the colored jones polynomial . in this paper , it will be used to prove the existence of the head and tail for all adequate links : [ skein ] if @xmath50 is a b - adequate diagram of the link @xmath1 , and @xmath113 is the corresponding graph , then @xmath114 this theorem was proved for the case when @xmath50 is a reduced alternating diagram in @xcite as theorem 4.3 .
the proof given there extends easily to b - adequate diagrams .
we will present the proof again here with the modifications .
in later sections , we shall denote @xmath115 . for a rational function @xmath61 ,
let @xmath116 be the minimum degree of @xmath61 considered as a power series .
the theorem will now follow from the following three lemmas .
[ gamma ] @xmath117 [ fuse ] @xmath118 [ graph ] if @xmath107 is the graph coming from a b - adequate diagram , then @xmath119 @xmath120 clearly @xmath121 increases as @xmath122 decreases . furthermore : @xmath123 to calculate @xmath124 note that in the previous formula for @xmath125 we get @xmath126 , @xmath127 , and @xmath128 . using this and the fact that @xmath129
, we get : @xmath130 consider the graph @xmath108 viewed as an element in @xmath131 .
we must compare @xmath132 with @xmath133 . recall that @xmath134 is @xmath135 times the number of circles in @xmath73 , where @xmath73 is obtained from @xmath53 by replacing the idempotents in the diagram by the identity in @xmath136 . for @xmath137 and @xmath138 ,
the number of circles in each diagram differ by @xmath75 .
thus @xmath139 . for @xmath113 , replacing the idempotents with the identitiy results in the all b - smoothing of the diagram @xmath140 .
since @xmath50 is a b - adequate diagram , so is @xmath140 . for @xmath141 , the replacement results in a smoothing of @xmath140 with exactly one a smoothing .
thus the result of the replacement for @xmath113 will have one more circle than the result of the replacement for @xmath141 , which give us @xmath142 . finally , since @xmath140 is b - adequate , @xmath113 is adequate .
thus by lemma [ replace ] , @xmath143
using theorem [ skein ] , the main theorem can be rewritten as follows : @xmath146 \draw ( 1,1.6 ) node[right]{\scriptsize{2n}}-- ( 1,3 ) ; \draw ( 1,3)--(0,4 ) node[left ] { \scriptsize{n } } ; \draw ( 1,3)--(2,4 ) node[right ] { \scriptsize{n } } ; \fill ( 1,3 ) circle ( 0.07 ) ; \end{tikzpicture } = \begin{tikzpicture}[baseline=2.8 cm ] \draw ( 1.7,3.5)--(1.7,4 ) node[right ] { \scriptsize{n } } ; \draw ( 0.3,3.5)--(0.3,4 ) node[left ] { \scriptsize{n } } ; \draw[ultra thick , fill = white ] ( 1.4,3.35 ) rectangle ( 2.0 , 3.5 ) ; \draw[ultra thick , fill = white ] ( 0,3.35 ) rectangle ( 0.6 , 3.5 ) ; \draw ( 0.45,3.35 ) .. controls ( 0.8,2.9 ) and ( 1.2,2.9 ) .. ( 1.55 , 3.35 ) ; \draw ( 1 , 3.2 ) node { \scriptsize{0 } } ; \draw ( 1,1.6 ) node[right]{\scriptsize{2n}}-- ( 1,2.1 ) ; \draw[ultra thick , fill = white ] ( 0.7,2.1 ) rectangle ( 1.3 , 2.25 ) ; \draw ( 0.85,2.25)-- ( 0.15 , 3.35 ) ; \draw ( 0.45 , 2.8 ) node[left ] { \scriptsize{n } } ; \draw(1.15,2.25)--(1.85 , 3.35 ) ; \draw ( 1.5,2.8 ) node[right ] { \scriptsize{n } } ; \end{tikzpicture } = \begin{tikzpicture}[baseline=2.8 cm ] \draw ( 1.7,3.5)--(1.7,4 ) node[right ] { \scriptsize{n } } ; \draw ( 0.3,3.5)--(0.3,4 ) node[left ] { \scriptsize{n } } ; \draw[ultra thick , fill = white ] ( 1.4,3.35 ) rectangle ( 2.0 , 3.5 ) ; \draw[ultra thick , fill = white ] ( 0,3.35 ) rectangle ( 0.6 , 3.5 ) ; \draw ( 1,1.6 ) node[right]{\scriptsize{2n}}-- ( 1,2.1 ) ; \draw[ultra thick , fill = white ] ( 0.7,2.1 ) rectangle ( 1.3 , 2.25 ) ; \draw ( 0.85,2.25)-- ( 0.15 , 3.35 ) ; \draw ( 0.45 , 2.8 ) node[left ] { \scriptsize{n } } ; \draw(1.15,2.25)--(1.85 , 3.35 ) ; \draw ( 1.5,2.8 ) node[right ] { \scriptsize{n } } ; \end{tikzpicture } = \begin{tikzpicture}[baseline=2.8 cm ] \draw ( 1,1.6 ) node[right]{\scriptsize{2n}}-- ( 1,3 ) ; \draw ( 1,3)--(0,4 ) node[left ] { \scriptsize{n } } ; \draw ( 1,3)--(2,4 ) node[right ] { \scriptsize{n } } ; \draw[ultra thick , fill = white ] ( 0.7,2.9 ) rectangle ( 1.3 , 3.1 ) ; \end{tikzpicture}\ ] ] in general , @xmath144 will reduce to a collection of circles coming from the all - b smoothing @xmath52 , `` fused '' together with the jones - wenzl idempotent colored @xmath147 for each maximal negative twist region .
we will call this reduced form @xmath148 .
we would now like to consider @xmath149 and show that we can reduce it to @xmath148 without affecting the lowest @xmath150 terms .
to do this we will first show a local relation which we will be able to use repeatedly .
[ skeinreduc ] @xmath151{skeinlemma1v2.pdf } } \raisebox{12pt}{$\ ; = \;$ } \includegraphics[width=1in]{skeinlemma2v2.pdf } \raisebox{12pt}{$\ ; - \left(\frac{\delta_{k-1}}{\delta_{n+k}}\right)\;$ } \raisebox{3pt}{\includegraphics[width=1in]{skeinlemma6v2.pdf}}\ ] ] using the recursive formula for the jones - wenzl idempotent on the left of the left hand side of the identity we get : @xmath151{skeinlemma1v2.pdf } } \raisebox{12pt}{$\ ; = \;$ } \includegraphics[width=1in]{skeinlemma2v2.pdf } \raisebox{12pt}{$\ ; - \left(\frac{\delta_{n+k-1}}{\delta_{n+k}}\right)\;$ } \raisebox{3pt}{\includegraphics[width=1in]{skeinlemma3v2.pdf}}\ ] ] applying the recursive formula again on the middle idempotent of the right most diagram we get : @xmath152{skeinlemma3v2.pdf } & \raisebox{8pt}{$=$ } & \includegraphics[width=1in]{skeinlemma4av2.pdf } \raisebox{8pt}{$\;- \left(\frac{\delta_{n+k-2}}{\delta_{n+k-1}}\right)\;$ } \includegraphics[width=1in]{skeinlemma4v2.pdf}\\ & \raisebox{8pt}{$=$ } & \raisebox{8pt}{$- \left(\frac{\delta_{n+k-2}}{\delta_{n+k-1}}\right)\;$ } \includegraphics[width=1in]{skeinlemma4v2.pdf}\\ & \raisebox{8pt}{$=$ } & \raisebox{8pt}{$- \left(\frac{\delta_{n+k-2}}{\delta_{n+k-1}}\right)\;$ } \includegraphics[width=1in]{skeinlemma5v2.pdf}\end{aligned}\ ] ] now when we apply this recursive formula again , the first term will again be zero , and we can continue this process until we get .
@xmath153{skeinlemma3v2.pdf } \raisebox{9pt}{$\;= - \left(\frac{\delta_{k-1}}{\delta_{n+k-1}}\right)\;$ } \includegraphics[width=1in]{skeinlemma6v2.pdf}\ ] ] consider a circle @xmath100 in @xmath52 .
the circle @xmath100 appears in @xmath149 , although it runs through several idempotents .
the goal of the argument is to remove one copy of the circle @xmath100 from the idempotents .
once this is done for each circle in @xmath52 , then @xmath149 will have been reduced to @xmath148 . because the diagram @xmath50 is alternating , the circle @xmath100 bounds a disk which does not contain any of the other circles in @xmath52 .
this means that in @xmath149 , the circle @xmath100 looks like figure [ loc ] . here
all of the arcs ar labeled @xmath154 .
apply lemma [ skeinreduc ] to get the following relation : @xmath155{sblocalskein1.pdf } \raisebox{42pt}{$\;=\;$ } \includegraphics[width=1.5in]{sblocalskein2.pdf } \raisebox{42pt}{$\;- \left(\frac{\delta_{k-1}}{\delta_{n+k}}\right)\;$ } \includegraphics[width=1.5in]{sblocalskein3.pdf}\ ] ] this argument will be applied to each circle in succession , so @xmath14 is either @xmath29 or @xmath154 depending on whether the argument has been applied to that circle yet .
all non - labelled arcs are either @xmath29 or @xmath154 .
now @xmath149 is expressed as the sum of two terms , and the claim is that the minimum degree of the second term is at least the minimum degree of the first term plus @xmath150 .
thus the equation simplifies to : @xmath155{sblocalskein1.pdf } \raisebox{42pt}{$\;\dot{=}_{4(n+1)}\;$ } \includegraphics[width=1.5in]{sblocalskein2.pdf}\ ] ] to see that the claim is true , first note that @xmath156 now we need to compare the degree of the two diagrams involved . by lemma [ replace ]
we can get a lower bound for the minimum degrees of these diagrams , and since the knot diagram that these came from was b - adequate , the first term will be an adequate diagram , and thus , the lower bound will be equal to the actual minimum degree . note that the element in @xmath157 shown in figure [ hhh ] can be expressed as @xmath158 . in the previous equation
, this element appears in the right most term .
when comparing this terms to the first term , each @xmath70 merges two circles into one circle .
thus the number of circles in the diagrams differ by @xmath154 . and ,
finally , a circle can be removed and replaced with a factor of @xmath82 .
this tells us that the difference in the minimum degrees of the diagrams is at least @xmath159 .
putting this together with the difference in degrees of the coefficients , the difference in minimum degrees of the terms themselves is at least @xmath150 .
@xmath161{sblocalfinal.pdf } \raisebox{42pt}{$\;=\;$ } \includegraphics[width=1.5in]{sblocalfinal2.pdf } \raisebox{42pt}{$\;= \left(\frac{\delta_{n+k+1}}{\delta_{n+k}}\right)\;$}\includegraphics[width=1.5in]{sblocalfinal3.pdf}$ ] this proves theorem [ restated ] in the case of reduced alternating diagrams . for the case
when the diagram @xmath50 is b - adequate , most of the proof still applies .
the only thing that goes wrong is that figure [ loc ] is not accurate because a circle @xmath100 in @xmath52 might not bound a disk , and thus in @xmath165 , may have idempotents which alternate which side of the circle it fuses to other circles .
figure [ adtref ] shows a non - alternating b - adequate diagram of the trefoil where the dotted cicle is an example such a circle @xmath100 .
we would still like to pull out one copy of @xmath100 , but in this case we must be more careful while doing so .
first we will modify the diagram as in figure [ twisting ] by adding crossings along the circle @xmath100 between any pair of idempotents which alternate wihch side of @xmath100 is the outer side .
the procedure is to modify the diagram so that the outer strand passes over all of the other copies of @xmath100 , so that it is still the outer strand when it meets the next idempotent in line .
call this new diagram @xmath84 . when expanding @xmath84 by summing over all possible smoothings of the crossings , only one state is non - zero , and that state is @xmath165 .
since this particular smoothing has an equal number of a and b smoothings , we get that @xmath166 .
now we will still apply the procedure to try to pull out one copy of @xmath100 as in the alternating case .
the argument still applies with no modification , as long as we note that in @xmath167 , there are still @xmath154 parallel copies of @xmath100 , and they are still unknotted because one of the copies is commpletely over the other copies .
thus that copy can be straightened out to lie next to the other copies .
when applying lemmma [ skeinreduc ] , the second term is still has @xmath154 fewer circles than @xmath167 because of the same argument .
finally we have the equation : @xmath168{bads.pdf } & \raisebox{24pt}{$=$ } & \includegraphics[height=.8in]{badt.pdf}\\ & \raisebox{24pt}{$=$ } & \includegraphics[height=.8in]{badtapplied.pdf}\\ & \raisebox{24pt}{$=$ } & \includegraphics[height=.8in]{badtalmost.pdf}\\ & \raisebox{22pt}{$=$ } & \raisebox{22pt}{$\left(\frac{\delta_{n+k+1}}{\delta_{n+k}}\right)$ } \includegraphics[height=.7in]{badtdonev2.pdf}\\ & \raisebox{22pt}{$\dot{=}_{4(n+1)}$ } & \includegraphics[height=.7in]{badtdonev2.pdf}\\\end{aligned}\ ] ] which completes the argument . | we investigate the coefficients of the highest and lowest terms ( also called the head and the tail ) of the colored jones polynomial and show that they stabilize for alternating links and for adequate links . to do this
we apply techniques from skein theory . |
the inclusive ( inelastic ) production of beam like particles , known as single diffraction , as in : @xmath11 and its analogous @xmath12 and ep interactions , presents one of the most interesting phenomena in strong interaction physics . an observed rapidity gap "
( absence of particles in a range of rapidity ) between @xmath13 and @xmath14 in the final state signifies that the entire ( color singlet ) residual momentum of the proton , with beam momentum fraction , @xmath15 , participates in the interaction between it and the second beam particle .
this effect is described in terms of the exchange of the regge trajectory@xcite , which embodies the idea of factorization " .
the momentum transfer , , and the beam momentum fraction , , of the final state proton , @xmath14 , `` tag '' the corresponding parameters of the exchanged . since @xmath16 is observed to be the most likely beam momentum fraction of the final
state @xmath14 , correspondingly the most likely value of the s momentum fraction in the proton , @xmath3 , is near zero . nonetheless , at current collider energies the squared invariant
mass of the system @xmath13 in eq .
[ eq : dif ] , @xmath17 to good approximation , can be quite large .
this fact led to a proposal@xcite to study hard scattering in such interactions , as a means of determining if the possesses an observable partonic structure .
the observation of the predicted hard scattering@xcite supported the notion that the behaves like a quasi real object inside the proton with an effective flux factor .
an open question is to what extent such a flux factor is universal ; for example , is it independent of beam particle or center - of - mass energy , or are there regions of phase space where factorization breaks down , due to interference with other more complex phenomena ( e.g. multiple -- exchange ) ?
one of the long standing theoretical problems in high energy hadronic interactions has been the understanding of @xmath6channel unitarization in exchange ( diffractive ) interactions .
empirically , one finds that the total diffractive cross section , , in reaction [ eq : dif ] and in the corresponding @xmath12 interaction , initially rises from threshold and tends to level off or flatten " at high energy@xcite , whereas the dominant triple@xcite description of these processes ( see below ) continues to rise and soon exceeds the total @xmath18 cross section .
there is no built in mechanism in the pure triple
process to account for the observed flattening of , and hence avoid the violation of unitarity .
figure 1 displays the problem [ 817 ] .
the @xmath6-dependence of the total cross section for react .
[ eq : dif ] , is shown , because the experimental acceptance usually depends weakly on in this region and because the integrated background from non- exchange and other sources in this region is small enough to be neglected . ] for feynman@xmath19 ( or @xmath20 ) of the final state proton or antiproton ( the domain where exchange is dominant )
. rises sharply from its threshold at 11.3 gev beam momentum and gently levels off to @xmath219 mb at the highest fermilab energy .
the solid curve in fig . [ fig : sigtot ] is the triple - regge prediction discussed below . at high energies , it is in complete disagreement with the measured cross section . in the continuing theoretical efforts to satisfy @xmath6channel unitarity [ 1822 ] , the words , screening , shadowing , absorption and damping are all used@xcite to describe effects due to multiple exchange ( two --
exchange is also an important component in understanding @xmath12 elastic scattering@xcite at low-@xmath4 ) .
these calculations have had varying degrees of success .
goulianos has taken a more pragmatic approach@xcite to satisfying unitarity and suggested that the integral of the flux factor in a proton should saturate at unity above @xmath22 gev . in the present letter
, we find that damping is confined to the low@xmath3 , low@xmath4 region .
we continue the analysis of the ua8 collaboration@xcite and demonstrate that there are regions either at larger @xmath3 or at larger , where the available data are well described by the triple - regge formula and therefore require no damping .
it is thus clear that the damping function depends on both @xmath3 and .
we attempt to determine the @xmath3dependence at low@xmath4 of an effective " multiplicative damping factor which could account for the discrepancies between data and solid curve in fig .
[ fig : sigtot ] .
however , we call it a toy damping factor " for several reasons . first , there are large gaps in the available data in fig .
[ fig : sigtot ] and some inconsistencies , therefore making it impossible to find a unique function .
secondly , the processes which give rise to the observed damping may imply a breakdown of factorization , in which case a simple universal damping factor may not exist at low@xmath3 and low@xmath4 .
finally , there is some evidence that the nature of the @xmath3dependence may itself depend on @xmath6 at our highest energies .
we briefly summarize the relevant formula .
the mueller regge expansion@xcite for the differential cross section of react .
[ eq : dif ] is : @xmath23 where @xmath24 is the regge trajectory for reggeon @xmath25 .
the sum is taken over all possible exchanged reggeons .
the @xmath26 are products of the various reggeon proton and triple
reggeon couplings and the signature factors .
there are two dominant terms in eq .
[ eq : tripler ] at small @xmath3 , namely @xmath27 and @xmath28 , where the first term corresponds to the triple process , and the second corresponds to other non leading , c=+ trajectories ( e.g. , @xmath29 ) in the proton interaction , the @xmath30 term increases with increasing @xmath31 , whereas the @xmath28 term decreases with increasing @xmath31 .
because the is the highest lying trajectory , when @xmath32 , @xmath33 is negative and the differential cross section increases sharply as @xmath34 .
this corresponds to the empirical observation that the most likely momentum fraction of the in the proton , @xmath3 , is near zero .
thus , the sharp rise in the triple regge prediction of in fig .
[ fig : sigtot ] is due to the kinematic fact that the minimum value of @xmath3 decreases with increasing @xmath6 as @xmath35 for fitting to data , eq .
[ eq : tripler ] has been rewritten@xcite as : @xmath36 \ , { \cdot}\ , \sigma_0 [ ( s')^{\epsilon_1 } \ , + \ , \rm r \ , \it ( s')^{\epsilon_2}],\ ] ] where , * the left hand bracket is taken as the flux factor , , and the right hand bracket ( together with the constant , @xmath37 ) is the proton total cross section , . * the two terms in correspond to the @xmath38 factor
, rescattering effects may lead to a logarithmic dependence on @xmath31 as well as to other complications@xcite . ] in the @xmath30 and @xmath28 terms in eq .
[ eq : tripler ] .
thus , has a form similar to that of real particle cross sections@xcite . *
the products @xmath39 and @xmath40 are , respectively , the values of @xmath41 and @xmath42 at = 0 .
* the trajectory , @xmath43 , has been shown@xcite to become relatively flat for @xmath1 gev@xmath2 ( see next section ) ; therefore a quadratic term is added to the standard linear trajectory@xcite , @xmath43 = 1.10 + 0.25 t + @xmath44 .
the non zero value of @xmath45 in @xmath46 compensates for the presence of the quadratic component in the trajectory factor . ] .
* @xmath47 is the standard donnachie
landshoff@xcite form factor . ] since it has never been shown to describe react .
[ eq : dif ] at large , the @xmath46 factor also serves as a possible correction .
thus , the product , @xmath48 , carries the dependence of the @xmath49 in eq .
[ eq : tripler ] and is assumed to be the same for both @xmath50 and @xmath51 .
physically , this means that the has the same flux factor in the proton , independent of whether the proton interaction proceeds via exchange or reggeon exchange .
we already know from the information in fig . [
fig : sigtot ] that the dominant contribution to the total cross section , namely the data with small@xmath3 and small@xmath4 , are not described by the triple regge formalism ; a damping of the flux with increasing @xmath6 is certainly required in this region .
however , we see no reason to suppose that the same damping must apply to the entire @xmath3 domain and contributes significantly to @xmath52 , as proposed by goulianos@xcite . however , this issue can be resolved by using available data to determine if there are regions in the @xmath3 plane where the formalism does apply ; that is , where damping is not required . the ua8 collaboration has recently reported@xcite a ( successful ) simultaneous fit of eq .
[ eq : triplep2 ] to their data on react .
[ eq : dif ] at @xmath53 gev and the extensive data sample of the chlm collaboration at the cern isr with @xmath54 = 23.5 and 30.5 gev .
they use the values , @xmath55 = 0.10 and -0.32 , obtained for the and @xmath56 trajectories , respectively , in fits to real particle total cross sections@xcite .
the four free parameters , @xmath57 , @xmath58 , are determined by fitting eq .
[ eq : triplep2 ] , plus an empirical background function of the form , @xmath59 , to the combined isr
ua8 data set in the range , there are no concerns about experimental resolution causing `` spill over '' from the large peak at @xmath60 . ] , @xmath610.10 .
the fitted parameters are : = @xmath57 = = = @xmath62 mb gev@xmath63 + @xmath64 = @xmath65 gev@xmath66 + @xmath45 = @xmath67 gev@xmath63 + @xmath68 = @xmath69 + this value of @xmath64 was independently confirmed@xcite in fits to the @xmath3dependence in the peak region with @xmath70 .
these triple regge results can be used to predict the total cross section of react .
[ eq : dif ] , , be integrating eq .
[ eq : triplep2 ] over the entire range , as well as for @xmath71 .
this yields the solid curve in fig .
[ fig : sigtot ] and illustrates the discrepancy@xcite with the experimental .
[ fig : dsdt035 ] from ref .
@xcite shows the isr@xcite and ua8 data@xcite data in the restricted region , @xmath72 , where the small ( @xmath2115% ) nonexchange background can be ignored ( the background is smaller than the size of the dots in the figure and about the same magnitude as the systematic uncertainty in absolute cross sections ) .
the similarity of at @xmath73 at both isr and energies reflects the fact that @xmath74 has nearly the same value at both @xmath75 = 5 and 118 gev . and in eq .
[ eq : triplep2 ] , is proportional to @xmath76 . ]
the term , @xmath77 , in eq .
[ eq : triplep2 ] makes this possible .
the solid curves in fig .
[ fig : dsdt035 ] are fits to these data _ without a background term _ and yield values of the 4 parameters which are in excellent agreement , @xmath78 , @xmath79 and @xmath80 , respectively . ] with those given above from the @xmath81 fit , thus lending credence to the relibility and stability of the fits .
we have found a second region in the @xmath3 plane , at small@xmath3 but large@xmath4 , where the triple regge formalism also describes the isr and ua8 data with no additional free parameters .
[ fig : sigtothit ] shows the high momentum transfer part of for @xmath71 and the limited @xmath4range , 1.02.0 gev@xmath2 , plotted vs. @xmath6 for the isr and ua8 data .
the solid curve is the prediction of eq .
[ eq : triplep2 ] using the above parameters .
the dashed curve is obtained by decreasing the @xmath45parameter by 1@xmath82 from its central value .
we see that , in contrast with the situation for , the triple regge formula , eq .
[ eq : triplep2 ] , accounts for the observed @xmath6dependence of the total single diffractive cross section in the high@xmath4 range , 1.02.0 gev@xmath2 .
the different shapes of the curves in figs .
[ fig : sigtot ] and the solid curve in fig .
[ fig : sigtothit ] are due to the dependence of the trajectory .
we have thus demonstrated that damping depends on both @xmath3 and ; it only exists in the small@xmath3 , small@xmath4 region , and is not required by data away from that region either at larger @xmath3 or at larger @xmath4 .
this could explain why cdf@xcite reports abnormally large backgrounds in triple fits to their ( low@xmath3 , low@xmath4 ) data at = 1800 gev . for example , at @xmath73 ( and @xmath4 = 0.05 gev@xmath2 ) where normally 1520% background is found , their fitted formula corresponds to nonexchange backgrounds of 51% .
such a result can be expected if a @xmath3dependent damping factor is required , but is left out of the fit ; the fitted ( large ) background term compensates for the wrong @xmath3dependence in the triple - regge equation without damping . since cdf only reports the fitted functions , we compare our prediction of with the sum of their signal " and background " terms .
this sum corresponds to the solid bands in fig .
[ fig : cdft ] ; the curves are the vs. predictions of eq .
[ eq : triplep2 ] at @xmath73 .
we see that , at 1800 gev , the prediction agrees to within @xmath83 of the cdf result ; at 546 gev , there is also good agreement at the lowest @xmath4value , although their fitted @xmath84-dependences at the two energies are not self - consistent .
the dependence of the disagreement between triple regge and the measured cross section is best seen by comparing the predictions with the experimental values of @xmath85 plotted vs. .
this is done in figs .
[ fig : dsdtall ] for eight isr energies and in fig .
[ fig : dsdtua8 ] at the collider and one at low@xmath86 which allowed them to span most of the available range . ] . the dashed and dotted curves in figs .
[ fig : dsdtall ] and [ fig : dsdtua8 ] are the ( undamped ) triple regge predictions for @xmath87 , calculated by integrating eq .
[ eq : triplep2 ] over the range , @xmath71 ( for the dotted curve , @xmath45 is decreased by 1@xmath82 from its central value ) . at the isr energies , where the triple regge prediction only exceeds the data by about 1015% ( see fig .
[ fig : sigtot ] ) , the differences between dashed curves and data in fig .
[ fig : dsdtall ] are hardly noticable , because the dot sizes are roughly similar to the discrepancies . at @xmath88 gev , however , the same effect is larger and highly visible . at that energy
, we see that there is a gradual transition from the low@xmath4 region which dominates , and where the experimental is smaller than the ( undamped ) triple regge prediction , to the higher@xmath4 region where the predictions agree with the data .
this seems to be a smooth transition over the @xmath4range , 0.51.0 gev@xmath2 .
the situation at the lower isr energies in figs .
[ fig : dsdtall ] is similar but less pronounced .
we conclude that the discrepancies between predictions and data are confined to the low@xmath4 region .
since , as noted above , the calculated rise in with increasing @xmath6 is due to the fact that , kinematically , @xmath89 decreases with increasing @xmath6 , it seems natural to introduce an empirical damping factor , @xmath90 , in eq .
[ eq : triplep2 ] which suppresses small @xmath3values ; _ i.e. , we strike at the `` heart '' of the problem_. d(@xmath3 ) will be unity everywhere except at small@xmath3 . for @xmath90
, we have tried a `` toy '' damping function which decreases from unity for @xmath91 , following a quadratic function as shown in fig .
[ fig : damp ] .
the parameters of the quadratic function are choosen to reproduce the leveling off of at isr energies in fig .
[ fig : sigtot ] . to account for the tevatron and points , an additional ( steep ) fall
off is needed for @xmath92 .
we arbitrarily use a cubic form .
@xmath90 has a total of 3 free parameters , since the quadratic and cubic have identical slopes and magnitudes at @xmath93 .
these were chosen to be the parameters of the quadratic function and the slope of the cubic function at @xmath94 . ] .
the dashed curves on fig .
[ fig : sigtot ] shows how such a function accounts reasonably well . ] for at high energies . the solid curves in figs .
[ fig : dsdtall ] and [ fig : dsdtua8 ] are calculated from eq .
[ eq : triplep2 ] multiplied by the above damping factor . as expected from fig .
[ fig : sigtot ] , the effect of damping is very small at isr energies , but increases with @xmath6 . at the energy of the
, however , the damping is about a factor of 3 in the low region , which dominates .
there is good agreement with the damping predictions at low@xmath4 in both figs .
[ fig : dsdtall ] and [ fig : dsdtua8 ] .
while , as explained above , the parameters of the quadratic term are chosen to agree with the departure of the isr cross sections from the triple - regge prediction in fig .
[ fig : sigtot ] , there seems to be no reason a - priori why this formulation should be valid at higher energies . to clarify this point
, we assume that the formula cdf fitted@xcite to their data is a sufficient description of and compare its @xmath3-dependence at @xmath4 = 0.05 gev@xmath2 with that of eq .
[ eq : triplep2 ] .
the band in fig . [ fig : damp ] shows the ratio of the cdf at 1800 gev to the triple - regge prescription , which can be interpreted as an empirical damping factor .
this decreases from near unity at @xmath61 to about 0.5 at @xmath95 , but is insufficient to account for the factor of @xmath215 required by the 1800 gev cross section in fig .
[ fig : sigtot ] ; therefore an additional ( rapid ) decrease in the damping factor must occur at smaller @xmath3 , analogous to our toy damping factor discussed above .
the cdf function thus indicates that , at larger @xmath6 , the onset of damping occurs at increasingly larger @xmath3values .
however , this does not invalidate the solid ( damped ) curve calculated with the above @xmath90 and shown in fig .
[ fig : dsdtua8 ] , because the overall damping calculation is not sensitive to the details of @xmath90 in the larger @xmath3-region .
we first summarize the key points of this letter : * we have a triple - regge parametrization of inclusive single diffraction which agrees with the data in two domains of the @xmath3 plane : ( a ) @xmath0 at all , ( b ) @xmath1 gev@xmath2 at all @xmath3 ( figs .
[ fig : dsdt035 ] , [ fig : sigtothit ] and [ fig : cdft ] ) .
since the triple - regge parametrization fails when applied to the full @xmath3 range of the total single diffractive cross section ( @xmath71 and all ) , we can conclude that damping occurs only at low@xmath3 and low@xmath4 .
* we have given a parametrization of the damping factor , d(@xmath3 ) , valid for @xmath96 gev@xmath2 , which describes all the low@xmath4 @xmath5 data at the isr and roughly accounts for the observed @xmath6dependence of up to tevatron energies ( figs .
[ fig : sigtot ] , [ fig : dsdtall ] , [ fig : dsdtua8 ] and [ fig : damp ] ) . * an effective damping factor calculated for the cdf fitted function for at @xmath8 gev and @xmath9 gev@xmath2 , suggests that , at fixed-@xmath3 , damping increases as @xmath6 increases ( fig .
[ fig : damp ] ) .
these results raise a number of issues : we can conjecture that , in the regions where the triple - regge formalism describes the data and there is no evidence of damping , factorization is valid and the -flux - factor may be universal . a systematic program of testing universality of the flux factor in these regions should be carried out in @xmath12 , @xmath97 and @xmath98 interactions .
our damping results may shed some light on the measurements of the @xmath99 quantity from the cross sections for diffractive jet production in react .
[ eq : dif ] and its analogue @xmath98 reaction ( k is the normalization constant in eq . [
eq : triplep2 ] and @xmath100 measures violation of the momentum sum rule , when @xmath101 ) .
assuming that the has dominant gluonic structure@xcite , there are three measurements of @xmath99 : [ cols= " < , < " , ] the fact that the ua8 data is at large-@xmath4 where there is no damping , whereas the cdf data is at small-@xmath4 where the damping factor in the region of the jets is of order 0.50 , could account for the difference between the ua8 and cdf @xmath99 values .
in addition , despite the large errors , it is interesting that the ua8 and zeus values for @xmath99 are consistent .
this might be expected , if there is no damping in @xmath98 collisions at high-@xmath102 .
of course , at low-@xmath102 , where the photon exhibits hadronic properties , multi- exchange , and hence damping , may result in smaller values of @xmath99 . in order to further study the possible @xmath6-dependence of the effective damping factor
, it would be very useful to make detailed measurements of single diffraction in pp interactions at rhic energies .
this would fill in the large gap getween isr and collider energies seen in fig .
[ fig : sigtot ] .
we note that ua8@xcite offers as possible explanations of their observed flattening of the trajectory for @xmath1 gev@xmath2 , either that it is an effect of multiple- exchange , or that it is evidence for the onset of the perturbative 2-gluon pomeron@xcite . in view of our observation that damping is not required in this -region , it seems that the perturbative , explanation is more likely
. it may therefore be interesting to study the @xmath103 cross section from @xmath104 up through the @xmath1 gev@xmath2 region .
we thank a. kaidalov , p. landshoff and e. gotsman for helpful comments . a.h .
mueller , phys .
d2 ( 1970 ) 2963 ; d4 ( 1971 ) 150 ; + a.b .
kaidalov et al . , pisma jetp 17 ( 1973 ) 626 ; + a. capella , phys .
d8 ( 1973 ) 2047 ; + r.d . field and g.c .
fox , nucl .
b80 ( 1974 ) 367 ; + d.p .
roy and r.g . roberts , nucl . phys .
b77 ( 1974 ) 240 . | we show that a triple - regge parametrization of inclusive single diffraction agrees with the data in the following two domains : ( a ) @xmath0 at all , ( b ) @xmath1 gev@xmath2 at all @xmath3 .
since the triple - regge parametrization fails when applied to the full @xmath3 range of the total single - diffractive cross section , we conclude that damping occurs only at low@xmath3 and low@xmath4 . we give a ( `` toy '' ) parametrization of the damping factor , d(@xmath3 ) , valid at low-@xmath4 , which describes the @xmath5 data at the isr and roughly accounts for the observed @xmath6dependence of up to tevatron energies .
however , an effective damping factor calculated for the cdf fitted function for @xmath7 at @xmath8 gev and @xmath9 gev@xmath2 , suggests that , at fixed-@xmath3 , damping increases as @xmath6 increases .
we conjecture that , in the regions where the triple - regge formalism describes the data and there is no evidence of damping , factorization is valid and the -flux - factor may be universal . with the assumption that the observed damping is due to multi- exchange , our results imply that the recent ua8 demonstration that the effective trajectory flattens for @xmath1 gev@xmath2 is evidence for the onset of the perturbative 2-gluon pomeron .
our damping results may also shed some light on the self - consistency of recent measurements of hard - diffractive jet production cross sections in the ua8 , cdf and zeus experiments .
= 0.75 mm = 0.75 mm 17 march , 1998 * * ' '' '' evidence for @xmath3 and -dependent damping + ' '' '' of the flux in the proton + samim erhan and peter e. schlein + university of california@xmath10 , los angeles , california 90095 , usa .
+ submitted to physics letters b + ' '' '' @xmath10 supported by u.s .
national science foundation grant phy9423142 + |
let @xmath2 be a complex manifold .
we denote by @xmath3 ( resp .
@xmath4 ) the sheaf of holomorphic functions ( resp . holomorphic differential operators ) on @xmath2 . for any reduced divisor @xmath5 in @xmath2 ,
we denote by @xmath6 $ ] the sheaf of meromorphic functions with poles along @xmath5 at most . given a regular holonomic @xmath4-module @xmath7 and a meromorphic function @xmath8 on @xmath2 with poles along @xmath5
, we define the @xmath4-module @xmath9 as being the @xmath3-module @xmath10:=\mathcal{o}_x[\ast z]\otimes_{\mathcal{o}_x}\mathcal{m}$ ] with connection @xmath11 defined as @xmath12 , where @xmath13 is the connection on @xmath10 $ ] given by its left @xmath4-module structure .
it is known that @xmath9 is a holonomic @xmath4-module .
it is irregular along @xmath5 .
we will say @xmath9 has _
exponential type_. in the algebraic setting , if @xmath14 is a smooth algebraic variety over @xmath15 , we use an exponent `` @xmath16 '' in the notation of the corresponding sheaves and we define the notion of a holonomic @xmath17-module of exponential type similarly . let @xmath9 be a holonomic @xmath4-module of exponential type .
the cohomology sheaves of its direct image by any proper map @xmath18 to a complex curve are holonomic @xmath19-modules with possibly irregular singularities .
the main result of this article consists in the computation of the formal invariants of these cohomology modules at their singularities .
the basic computation takes place in the following setting : the complex manifold @xmath2 is the product @xmath20 of a small disc @xmath21 centered at the origin in @xmath15 by the riemann sphere , the meromorphic function @xmath8 is the projection @xmath22 with polar divisor @xmath23 and the map @xmath24 is the projection @xmath25
. we will denote by @xmath26 a coordinate on @xmath21 .
as above , @xmath7 is a regular holonomic @xmath4-module .
the only interesting cohomology module to consider is @xmath27 and we assume that @xmath21 is small enough so that @xmath28 is the only singular point of @xmath29 . in a neighbourhood @xmath14 of @xmath30 , the singular support of @xmath7 is an analytic curve @xmath31 .
we denote by @xmath32 the set consisting of the local irreducible components of @xmath31 which are distinct of @xmath33 or @xmath34 .
set @xmath35 .
to any @xmath36 at @xmath37 we associate : * @xmath38 : the multiplicity of the conormal space @xmath39 in the characteristic cycle of @xmath7 .
* @xmath40 : the intersection multiplicity at @xmath37 of @xmath36 with @xmath41 and @xmath42 , respectively .
* @xmath43 : the convex hull in @xmath44 of the union of @xmath45 and @xmath46 . *
@xmath47 $ ] and @xmath48 : the polar part and the holomorphic part of a puiseux parametrization of @xmath36 at @xmath37 .
let @xmath49 , @xmath50 , such that @xmath51 is a parametrization of @xmath36 .
@xmath52 is the polar part of @xmath53 and @xmath54 is its holomorphic part ( @xmath55 ) .
the irregularity number of @xmath56 is computed in [ 9 ] .
but , we can go further in computing formal invariants of @xmath56 . [ 1 ] @xmath57 after a suitable translation , the newton polygon of the @xmath58-module @xmath56 is the minkowski sum @xmath59 . in particular ,
the set of slopes of @xmath56 is @xmath60 and the irregularity number of @xmath56 is @xmath61 .
@xmath62 let @xmath63 .
after the ramification @xmath64 , @xmath65 , the formal irregular part of @xmath66 decomposes as @xmath67 where : * @xmath68 $ ] is a finite subset and @xmath69 if and only if there exist @xmath70 and @xmath71 with @xmath72 such that @xmath73 the set of such @xmath70 is denoted by @xmath74 . *
@xmath75 is a regular holonomic @xmath76\langle\partial_\tau\rangle$]-module , @xmath77\neq0 $ ] and the rank of @xmath75 is @xmath78 .
but we can also compute the characteristic polynomial of the monodromy of @xmath79 under the assumption @xmath80 : for any @xmath81 and for any @xmath82 with @xmath83 and @xmath84 , @xmath85 we remark that @xmath86 is a normalization of all the @xmath87 s . for any @xmath70 , @xmath88 is the union of some smooth irreductible analytic curves @xmath89 .
then to any @xmath87 we associate @xmath90 $ ] : the characteristic polynomial of the monodromy of @xmath91 around @xmath92 in a normalization of @xmath87 , where @xmath93 is an equation of @xmath87 . here
, @xmath94 is the vanishing cycle functor along @xmath87 for bounded complex with constructible cohomology ( cf .
@xcite , @xmath95 ) and @xmath96 is the de rham functor ( cf . definition @xmath97 p. @xmath98 in @xcite ) .
@xmath91 has support included in @xmath87 and @xmath99 is just a local system .
then , @xmath100 is the characteristic polynomial of the monodromy of the local system @xmath101 around @xmath92 ( do not confuse with the monodromy around @xmath87 ) . as @xmath102 , for @xmath103 , @xmath100 is independant of @xmath104 .
[ zeta ] under the assumption @xmath80 , the characteristic polynomial of the monodromy of @xmath75 is equal to @xmath105 .
conversely , we prove in theorem [ realisation ] that given any formal holonomic @xmath1\langle\partial_t\rangle$]-module @xmath106 , there exist a ramification @xmath107 and a regular holonomic @xmath108-module @xmath7 , such that @xmath109 is isomorphic to the formalization of @xmath110 . in the algebraic setting ,
let us assume that @xmath14 is affine and that @xmath24 and @xmath8 are regular functions on @xmath14 .
let @xmath7 be a holonomic @xmath111-module which has regular singularities included at infinity .
the direct images @xmath112 are holonomic @xmath113\langle\partial_t\rangle$]-modules with singularities at finite distance on the affine line @xmath114 and at infinity .
we will reduce the computation of the formal invariants of @xmath115 at each of its singularities @xmath116 to the situation of theorem [ 1 ] through the diagram @xmath117 let @xmath118 denote the direct image @xmath119 .
it consists in a complex of @xmath120-modules with regular holonomic cohomology modules @xmath121 .
[ 2 ] the germs @xmath122 and @xmath123 have the same formal irregular part . + singular support of @xmath124 .
@xmath125 is a regular singularity of @xmath115 .
@xmath126 and @xmath127 are irregular singularities .
let @xmath14 be a smooth affine surface .
suppose @xmath128 are algebraically independant and @xmath129 .
the only interesting cohomology module to consider is @xmath130 ; the others cohomology modules have punctual support ( cf .
proposition 5.1 of @xcite ) .
let @xmath116 and @xmath31 be the singular support of @xmath131 in the neighbourhood of @xmath132 .
theorem [ 1 ] and theorem [ 2 ] allow us to describe the newton polygon of @xmath133 , up to a suitable translation , using the set @xmath134 of irreducible components of @xmath31 which are distinct of @xmath135 or @xmath136 and the multiplicity @xmath137 of @xmath138 in the characteristic cycle of @xmath131 .
but we can give a geometrical interpretation of the @xmath87 s and the @xmath137 s .
let @xmath139 be a smooth compactification of @xmath14 such that there exists two holomorphic functions @xmath140 which extend @xmath24 and @xmath8 .
let @xmath141 be the critical locus of @xmath142 and @xmath21 be the divisor @xmath143 .
we denote by @xmath144 ( resp .
@xmath145 ) the cycle in @xmath146 which is the closure of @xmath147 ( resp .
@xmath148 ) , where the image is counted with multiplicity .
if @xmath149 ( resp .
@xmath150 ) , the @xmath87 s are the irreducible components of the germ @xmath151 ( resp .
@xmath152 ) at @xmath132 ; @xmath137 is the multiplicity of @xmath87 in @xmath151 ( resp .
@xmath152 ) . this computation can be found in @xcite ( theorem 6.4.1 ) .
we shall assume that the reader is familiar with the definitions and the properties of the specialization functor and the nearby cycles functor for @xmath153-modules along a hypersurface and we refer to @xcite , @xcite , @xcite , @xcite and @xcite for details .
given @xmath154 a smooth hypersurface of @xmath2 defined by a global equation @xmath155 and given @xmath7 a specializable @xmath4-module , we will denote by @xmath156 the specialization of @xmath7 along @xmath157 and by @xmath158 the nearby cycle module of @xmath7 along @xmath157 . if @xmath159 is the canonical v - filtration with respect to the lexical order on @xmath160 , @xmath161 and @xmath162 .
same definitions hold for @xmath157 hypersurface with normal crossings .
we will also denote by @xmath163 the nearby cycle functor for bounded complexes with constructible cohomology ( cf @xcite 1 ) .
let us explain the method we use to compute the formal invariants .
we note that if a summand @xmath164 shows up in the formal decomposition of a @xmath165-module @xmath166 after a ramification @xmath167 , then @xmath75 is the formal regular part of @xmath168 .
this formal regular part can be recovered from the specialization of @xmath169 at the origin ( cf .
example 5.2.1 in @xcite ) .
moreover , as @xmath79 is localized at the origin , the rank of @xmath79 and the characteristic polynomial of the monodromy of @xmath79 can be computed using the nearby cycles module of @xmath169 at the origin .
one of the main results we use is the commutation of the specialization functor ( resp .
nearby cycles functor ) with proper direct images ( cf .
theorem @xmath170 in @xcite for complexes of @xmath0-modules and theorem @xmath171 in @xcite for @xmath0-modules ) .
this property reduces the proof of theorem [ 1 ] to local analytic computations on a suitable blowing - up space of @xmath172 above @xmath37 .
it is also the main argument in the proof of theorem [ 2 ] .
this section is devoted to the proof of theorems [ 1 ] and [ zeta ] .
we first note that theorem [ 1 ] @xmath62 induces the construction of the newton polygon ( theorem [ 1 ] @xmath57 ) .
let @xmath173 .
denote by @xmath174 the set of @xmath69 with pole of order @xmath175 and by @xmath176 the set of @xmath70 such that @xmath177 .
the newton polygon of @xmath178 has just one slope , equal to @xmath175 ; its height is the product of the rank of @xmath179 by the slope . according to theorem [ 1 ] @xmath62 ,
the slope is equal to @xmath180 , for all @xmath181 , and the height is @xmath182 .
if @xmath183 denotes the convex hull in @xmath44 of the union of @xmath184 and @xmath185 , the newton polygon of @xmath186 is , after a suitable translation , the minkowski sum @xmath187 .
we deduce the newton polygon of @xmath166 from the one of @xmath186 by a dilation of the vertical axis in a ratio @xmath188 ( cf .
lemma 5.4.3 p. @xmath189 of @xcite ) .
we focus now on the proof of theorem [ 1 ] @xmath62 and theorem [ zeta ] .
the choice of the ramification enables us to reduce the proof of theorem [ 1 ] and [ zeta ] to the case where the @xmath190 s are equal to @xmath191 ; it is the non ramified case ( @xmath192 ) .
we begin with a base change formula .
let @xmath193 and @xmath194 be the canonical projections .
consider the cartesian diagram : @xmath195^{\rho^{'}}\ar[d]^{p_1^{'}}&d\times\mathbb{p}^1\ar[d]^{p_1}\\ d^{'}\ar[r]^{\rho}&d.}\ ] ] [ comm ] @xmath196 .
denote by @xmath197 ( resp .
@xmath198 ) the relative de rham functor of @xmath199 ( resp .
@xmath200 ) .
we adopt the convention that the relative de rham complexes are concentrated in negative degrees .
we have isomorphisms of complexes of @xmath201-modules : @xmath202 we deduce that @xmath203
. then we remark that the set @xmath204 consists of the irreducible components of the singular support @xmath205 of @xmath206 in a neighbourhood @xmath207 of @xmath37 , which are distinct of @xmath33 or @xmath208 .
let @xmath209 be the @xmath210-roots of the unity .
up to change of indices @xmath104 , a parametrization of @xmath89 is given by @xmath211 ; in particular , the intersection multiplicity @xmath212 at @xmath37 of @xmath89 with @xmath33 is @xmath191 . in the next section
, we will prove theorem [ 1 ] @xmath62 and theorem [ zeta ] in the case where all the @xmath210 s are equal to @xmath191 .
then , we may apply it to @xmath206 . at last
, we have to compute the data associated with each @xmath89 using data associated with the @xmath87 s .
but we have : + @xmath213 the polar part of a puiseux parametrization of @xmath89 at @xmath37 is @xmath214 .
moreover , if the assumption @xmath80 is fulfilled for @xmath7 it is also fulfilled for @xmath206 , + @xmath213 @xmath137 is the multiplicity of @xmath215 in the characterstic cycle of @xmath206 , + @xmath213 @xmath100 is the characteristic polynomial of the monodromy of the local system @xmath216 around @xmath92 , where @xmath217 is an equation of @xmath89 . if we prove that the rank of @xmath79 is equal to @xmath219 , the irregularity number at @xmath28 of @xmath29 and of @xmath220 are both equal to @xmath221 ( cf .
@xcite for @xmath29 ) .
then it is sufficient to prove that each summand @xmath222 , with @xmath69 , shows up in the formal decomposition of @xmath166 and to compute the rank and the characteristic polynomial of the monodromy of each @xmath79 .
let @xmath69 .
we denote by @xmath223 the specialization of @xmath224 at the origin .
@xmath225 is a @xmath226\langle\partial_t\rangle$]-module . as announced in the introduction
, @xmath79 is the formal regular part of @xmath227 .
it can be recovered using the isomorphism @xmath228\otimes_{\mathbb{c}[t]}\operatorname{sp}_0(\mathcal{n}e^{-\alpha})_0 $ ] ( cf .
example @xmath229 of @xcite ) .
let @xmath230 be the nearby cycles module of @xmath224 at the origin .
@xmath231 is a finite dimensional @xmath15-vector space equipped with an endomorphism of monodromy induced by @xmath232 .
as @xmath224 is localized at the origin , the rank of @xmath79 is the dimension of @xmath231 and the characteristic polynomials of their monodromies are equal .
but we have , @xmath233)$ ] . indeed , @xmath234e^{-\alpha}\otimes_{\mathcal{o}_d}\mathcal{h}^0p_{1+}(\mathcal{m}e^{p_2}),\\ & = \mathcal{o}_d[\frac{1}{t}]e^{-\alpha}\otimes^{\mathbb{l}}_{\mathcal{o}_d}\mathcal{h}^0p_{1+}(\mathcal{m}e^{p_2}),\text { ( flatness),}\\ & = ( \mathcal{o}_d[\frac{1}{t}]e^{-\alpha}\otimes_{\mathcal{o}_d}\mathcal{d}_d)\otimes^{\mathbb{l}}_{\mathcal{d}_d}\mathcal{h}^0p_{1+}(\mathcal{m}e^{p_2}),\\ & = \mathcal{h}^0p_{1+}((\mathcal{o}_{d\times\mathbb{p}^1}[\frac{1}{t}]e^{-\alpha\circ p_1}\otimes_{\mathcal{o}_{d\times\mathbb{p}^1}}\mathcal{d}_{d\times\mathbb{p}^1})\otimes_{\mathcal{d}_d}^{\mathbb{l}}\mathcal{m}e^{p_2}),\\ & \text { ( prop 2.6.6 in \cite{kasc}),}\\ & = \mathcal{h}^0p_{1+}(\mathcal{m}e^{p_2-\alpha\circ p_1 } ) .
\end{array}\ ] ] let @xmath235 .
according to the commutation of the nearby cycle functor with proper direct image ( cf .
theorem 4.8 - 1 p. @xmath236 of @xcite ) , @xmath237 ) , \text { ( cf . \cite{ma3 } p.$5 $ ) . }
\end{array}\ ] ] then , theorem [ 1 ] @xmath62 and theorem [ zeta ] follows from the proposition : [ 11 ] @xmath238 has support included in @xmath37 and the euler characteristic of @xmath239 is @xmath240 . + moreover , under the assumption @xmath80 , the zeta function of its monodromy is @xmath241 .
the proof is given in the next sections .
in the non ramified case , the assumption @xmath80 is equivalent to : @xmath242 in this section , we prepare the proof of proposition [ 11 ] by giving some computations of nearby cycles modules along a normal crossing .
let @xmath243 be a neighbourhood of @xmath92 in @xmath244 and @xmath245 be some coordinates on @xmath243 .
let @xmath7 be a regular holonomic @xmath246-module and @xmath247 .
we are interested in nearby cycles modules of the type @xmath248 .
suppose that the singular support of @xmath7 is included in @xmath250 .
let @xmath251 .
@xmath252 is a holomorphic connection ; we denote by @xmath253 its rank .
thus @xmath254 is a local system . for @xmath255
, we consider the monodromy @xmath256 of @xmath257 around @xmath258 .
as @xmath259 is connected and the singular support of @xmath7 is included in a normal crossing , the characteristic polynomial of @xmath256 does not depend on @xmath260 ( cf .
i.2.2 p. 55 in @xcite ) .
we denote it by @xmath261 .
[ b ] 1 .
if @xmath262 , @xmath263 .
2 . if @xmath264 and @xmath265 , @xmath266 .
3 . if @xmath267 , the euler characteristic of @xmath268e^{{{1}/{v}}})_{(0,0)}$ ] is equal to @xmath269 and the zeta function of its monodromy is @xmath270 .
the proof of the last two points is given in lemma @xmath271 of @xcite and its proof .
the shift in the third point comes from the convention on the de rham complexes . the first point is also proved in @xcite in the case where @xmath272 $ ] . to conclude ,
it is sufficient to prove it in the case where @xmath7 has support in @xmath273 .
we have @xmath274 $ ] , where @xmath275 , and @xmath276 $ ] is just a module . as @xmath277e^{{{1}/{u^k}}}))=0 $ ] ( computation in one variable ) , we deduce that @xmath263 .
let @xmath278 be the singular support of @xmath7 and @xmath279 be the set consisting of the irreducible components of @xmath278 which are distinct of @xmath258 .
we will assume that * the intersection multiplicity at @xmath92 of @xmath280 with @xmath258 is equal to @xmath191*. we denote by @xmath253 the rank of @xmath7 .
to any @xmath280 , we associate @xmath137 , the multiplicity of the conormal space @xmath281 in the characteristic cycle of @xmath7 . [
a ] the euler characteristic of @xmath282)_{(0,0)}$ ] is equal to @xmath283 . as @xmath7 is regular holonomic , @xmath284)&=\operatorname{dr}~\psi_u(\mathcal{m})\text { ( proposition 2.4 - 3 in \cite{memai}),}\\ & = \psi_u(\operatorname{dr}~\mathcal{m})\text { ( theorem $ 4.10.1 $ p. 233
in \cite{mesa}).}\end{array}\ ] ] let @xmath285 and @xmath286 small enough such that any @xmath280 intersects @xmath287 at a unique point @xmath288 ( by assumption ) and @xmath289 ( cf .
remark @xmath290 in @xcite ) .
denote by @xmath291 .
let @xmath292 , @xmath70 , be some disjoint balls centered at @xmath288 of radius small enough . according to the index theorem of kashiwara ( theorem 6.3.1 in @xcite ) , the euler characteristic of @xmath257 , @xmath293 , ( resp .
@xmath294 ) is equal to @xmath253 ( resp .
@xmath295 ) . as @xmath296 is constructible with respect to the stratification @xmath297 , the euler characteristic of @xmath298
may be computed using the mayer - vietoris theorem .
we obtain : @xmath299 the zeta function of the monodromy of @xmath282)_{(0,0)}$ ] can not be compute with this method in the general case .
nevertheless , we can do it when the singular support of @xmath7 is included in the normal crossing @xmath250 . we adopt the notations of the section [ nota ] .
let @xmath300 .
we denote by @xmath301 the characteristic polynomial of the monodromy of the local system @xmath302 around @xmath92 .
[ c ] if the singular support of @xmath7 is included in a normal crossing @xmath250 , the zeta function of the monodromy of @xmath282)_{(0,0)}$ ] is equal to @xmath303 .
we have @xmath282)_{(0,0)}=\underset{\underset{\epsilon,\eta}{\leftarrow}}{\lim}r\gamma(x_{\epsilon,\eta},\operatorname{dr}~\mathcal{m})$ ] . as the singular support of @xmath7 is included in @xmath250 and as @xmath304 is constructible with respect to a stratification such that @xmath250 is a union of strata
, we have @xmath305 , where @xmath306 is the intersection point of @xmath307 with @xmath273 .
then , the zeta function of the monodromy of @xmath282)_{(0,0)}$ ] coincide with the zeta function of the monodromy of the complex of local systems @xmath308 around @xmath92 .
consider the triangle of complexes of local systems : @xmath309 as the zeta function of the monodromy of @xmath310 around @xmath92 is @xmath261 , lemma [ c ] follows .
+ first we have to prove that @xmath311 , for any @xmath312 .
but in a neighbourhood @xmath14 of @xmath313 in @xmath314 , there exist coordinates @xmath315 such that @xmath316 , @xmath317 and @xmath318 , for some @xmath319 . therefore , @xmath320 .
as the singular support of @xmath321 is not necessarily a normal crossing at @xmath92 , we can not apply directly lemma [ b ] 1 .
the idea is to use a resolution @xmath322 , with exceptional locus @xmath323 , such that the singular support of @xmath324 $ ] has normal crossings in a neighbourhood of any point of @xmath323 .
as @xmath7 is holonomic and @xmath325 is a proper isomorphism out of @xmath323 , @xmath326 ( cf .
proposition 7.4.5 of @xcite ) . using the commutation of the proper direct image with @xmath327 and @xmath96 ( theorem 4.8.1 p. 226 in @xcite and
theorem 5.4.3 p. 77 in @xcite ) , we deduce that [ 9 ] @xmath328 * if @xmath260 is a smooth point of @xmath329 , we have @xmath330 ( cf .
lemma [ b ] 1 ) . *
if @xmath260 is a normal crossing point of @xmath329 , according to lemma [ b ] 2 , @xmath330 . then @xmath331 and @xmath238 has support included in @xmath37 .
let us now focus on @xmath332 . in a neighbourhood @xmath14 of @xmath37 in @xmath314
, there exist coordinates @xmath315 such that @xmath316 , @xmath333 and @xmath334 . to simplify notation
, we will set @xmath335 .
then we have : @xmath336e^{g})_{(0,0)}.\ ] ] we have to compute the euler characteristic and the zeta function of the monodromy of @xmath337e^g)_{(0,0)}$ ] .
we can not use directly the local computations because @xmath8 is not equivalent to a function of the type @xmath338 at @xmath92 .
but we will be led to this situation with the help of a resolution .
the choice of a good resolution is relevant for the end of the proof .
[ resolution ] there exists a resolution @xmath322 with exceptional locus @xmath323 ( actually it is a finite composition of the blow - up of the point @xmath339 and some blow - up of points in its exceptional locus ) which satisfies the following conditions : 1 .
let @xmath340 .
there exists a unique irreducible component @xmath341 of @xmath323 which intersects @xmath342 in a unique point @xmath260 .
moreover , in a neighbourhood of any point @xmath343 , we can choose local coordinates @xmath245 on @xmath139 such that : 1 . if @xmath344 , * @xmath345 and @xmath346 , with @xmath347 , + or * @xmath348 and @xmath349 , with @xmath350 , @xmath265 .
if @xmath351 , * @xmath258 ( resp .
@xmath273 ) is an equation of @xmath341 ( resp . of the other component of @xmath323 ) , * @xmath352 and @xmath353 , with @xmath267 .
3 . if @xmath354 , * @xmath258 is an equation of @xmath341 , * @xmath355 and @xmath356 , with @xmath357 .
the singular support of @xmath358)=\pi^{\ast}(\mathcal{m}_{|u})[\ast\widetilde{e}]$ ] has at most normal crossings at any point of @xmath359 .
, height=113 ] by reducing to the same denominator , we have @xmath360 , with @xmath361 $ ] and @xmath362 .
let us begin by proving the first point .
we will perform @xmath363 blow - up of points .
the resolution tree we will obtain has just one branch and the component @xmath341 will be the exceptional locus of the last blow - up .
the first @xmath364 blow - up allow us to decrease the degree on @xmath365 in the numerator .
we obtain a resolution @xmath366 composition of @xmath364 blow - up @xmath367 with exceptional locus @xmath368 , such that @xmath369 can be written locally as in the case ( a ) except at a point @xmath370 . in a neighbourhood of this point
, we can choose local coordinates @xmath245 such that @xmath355 and @xmath371 , with @xmath372 and @xmath373 .
now the @xmath374 next blow - up allow us to decrease the degree in @xmath375 in the denominator and the degree in @xmath375 of @xmath376 .
we obtain a new resolution @xmath366 , composition of @xmath377 blow - up , such that @xmath369 can be written locally as in the case ( a ) except at a point @xmath378 . in some
suitable local coordinates @xmath245 in a neighbourhood of @xmath379 , @xmath355 and @xmath380 , with @xmath126 constant and @xmath381 invertible . at last , we blow - up @xmath379 .
the exceptional locus of this last blow up is @xmath341 and we can verify the local behaviour of @xmath8 on @xmath341 .
the details are left to the reader . to prove the second point , it is sufficient to perform additional blow - up of points in @xmath359 .
we denote by @xmath280 the strict transform of @xmath87 .
[ stricttransform ] @xmath57 let @xmath70 .
@xmath382 if and only if @xmath383 .
@xmath62 the assumption @xmath80 is equivalent to : @xmath384 .
@xmath57 according to the choice of the resolution , @xmath382 if and only if @xmath385 .
let @xmath386 and @xmath245 be some local coordinates in the neighbourhood of @xmath387 such that @xmath388 and @xmath355 , with @xmath357
. as the intersection multiplicity of @xmath87 with @xmath33 is @xmath191 and as @xmath355 , the intersection multiplicity of @xmath280 with @xmath341 is also @xmath191 .
then : @xmath389 @xmath62 here , we want to understand the behaviour of @xmath87 after the resolution @xmath325 . we will keep the same notations as in the proof of lemma [ resolution ] .
let @xmath383 .
then @xmath87 has equation @xmath390 .
after the first @xmath364 blow - up , in the neighbourhood of the point @xmath370 with local coordinates @xmath245 , @xmath391 after the next @xmath392 blow - up , in the neighbourhood of the point @xmath379 , we can choose local coordinates @xmath245 such that @xmath393 where @xmath394 and @xmath126 is a constant which depends only on @xmath395 . at last , we blow - up @xmath379 and obtain that in the chart @xmath396 , @xmath397 , @xmath398 then , we see that the strict transform @xmath280 of @xmath87 interesct @xmath341 at a point which depends only on @xmath395 and @xmath399 .
the assertion @xmath62 follows .
there exists an isomorphism : @xmath400e^g)_{(0,0 ) } & = r\gamma(e,\operatorname{dr}~\psi_{\pi_1}(\pi^{\ast}(\mathcal{m}_{|u}[\frac{1}{xy}])e^{g\circ\pi})),\\ & \text{(same proof as for the isomorphism $ ( \ast\ast)$ p. \pageref{9}),}\\ & = r\gamma(e,\operatorname{dr}~\psi_{\pi_1}(\pi^{\ast}(\mathcal{m}_{|u})[\ast\widetilde{e}]e^{g\circ\pi } ) ) .
\end{array}\ ] ] let us first prove that the complex @xmath401e^{g\circ\pi})_{|e}$ ] has support included in @xmath341 .
we have two cases to consider : 1 . in a neighbourhood @xmath402 of a smooth point @xmath184 in @xmath403 , according to lemma [ resolution ] , there exist coordinates @xmath245 on @xmath243 such that @xmath345 , @xmath346 and the singular support of @xmath404 $ ] is included in @xmath250 . then , according to lemma [ b ] 1 , @xmath405e^{g\circ\pi})_q=\psi_{u^m}(\pi^{\ast}(\mathcal{m}_{|u})[\ast\widetilde{e}]e^{\frac{1}{u^k}})_{(0,0)}=0.\ ] ] 2 . in a neighbourhood @xmath402 of a normal crossing point @xmath184 in @xmath403 , according to lemma [ resolution ] , there exist coordinates @xmath245 on @xmath243 such that @xmath348 , @xmath349 and the singular support of @xmath404 $ ] is included in @xmath250 . then , according to lemma [ b ] 2 , @xmath405e^{g\circ\pi})_q=\psi_{u^mv^n}(\pi^{\ast}(\mathcal{m}_{|u})[\ast\widetilde{e}]e^{\frac{1}{u^kv^l}})_{(0,0)}=0.\ ] ] then @xmath337e^g)_{(0,0)}=r\gamma(e_d,\operatorname{dr}~\psi_{\pi_1}(\pi^{\ast}(\mathcal{m}_{|u})[\ast\widetilde{e}]e^{g\circ\pi}))$ ] .
now we want to examine the complex @xmath401e^{g\circ\pi})$ ] locally on @xmath341 .
let @xmath406 be a tubular neighbourhood of @xmath341 . according to corollary [ stricttransform ] ( 1 ) , the singular support @xmath407 of @xmath408)_{|t}$ ] is @xmath409 .
* computation of the euler characteristic : * we denote by @xmath410 the intersection point of @xmath411 with @xmath341 .
let @xmath253 be the rank of @xmath7 .
as @xmath325 is an isomorphism out of @xmath323 , the rank of @xmath412 is @xmath253 and the multiplicity of the conormal space @xmath413 in the characteristic cycle of @xmath412 is @xmath414 . 1 . in a neighbourhood @xmath402 of the point @xmath260 ,
according to lemma [ resolution ] , there exist coordinates @xmath245 on @xmath243 such that @xmath352 , @xmath353 and the singular support of @xmath404 $ ] is included in @xmath250 .
then , @xmath405e^{g\circ\pi})_p=\psi_{uv^n}(\pi^{\ast}(\mathcal{m}_{|u})[\frac{1}{uv}]e^{\frac{1}{v}})_{(0,0)}.\ ] ] + applying lemma [ b ] 3 .
, we conclude that the euler characteristic of the complex @xmath401e^{g\circ\pi})_p$ ] is equal to @xmath269 .
2 . in a neighbourhood @xmath402 of a point @xmath415
, @xmath369 is holomorphic and @xmath329 is smooth .
then we have @xmath401e^{g\circ\pi})_{q_i}=\operatorname{dr}~\psi_u(\pi^{\ast}(\mathcal{m}_{|u})[\frac{1}{u}])_{(0,0)}$ ] and , according to lemma [ a ] , the euler characteristic of the complex @xmath401e^{g\circ\pi})_{q_i}$ ] is @xmath416 .
3 . in a neighbourhood of a point @xmath417 , as in the previous point , we prove that the euler characteristic of the complex @xmath401e^{g\circ\pi})_q$ ] is @xmath253 ( because @xmath418 ) . using the mayer - vietoris theorem , we deduce that the euler characteristic of the complex @xmath419e^{g\circ\pi}))$ ] is equal to @xmath420 * computation of the zeta function of the monodromy : * assume that @xmath80 is fulfilled . the fact that for any @xmath70 , @xmath421 and the assumption @xmath80 imply that @xmath407 is a divisor with normal crossing ( cf .
corollary [ stricttransform ] ( 2 ) ) .
we denote by @xmath422 the complement of @xmath407 in @xmath406 and by @xmath387 the intersection point of one @xmath423 with @xmath341 , @xmath383 .
* @xmath424)_{|t^{\ast}}$ ] is a local system .
we can define the monodromy endomorphism around @xmath341 , @xmath425)_{p}\to ( \operatorname{dr}~\pi^{\ast}(\mathcal{m}_{|u})[\ast\widetilde{e}])_{p}$ ] , for any @xmath426 . as @xmath422 is connected and @xmath407 has normal crossings , the characteristic polynomial of @xmath256 does not depend on @xmath426 .
we denote it by @xmath261 . * to any @xmath383 , we associate the characteristic polynomial @xmath427 of the monodromy of the local system @xmath428 around @xmath92 , where @xmath429 is an equation of @xmath423 .
if @xmath430 is an equation of @xmath36 , as @xmath325 is an isomorphism out of @xmath323 , it coincides with the characteristic polynomial of the monodromy of @xmath431 around @xmath92 . using the local behaviour of @xmath432 and @xmath369 on @xmath341 given by lemma [ resolution ]
, we have : 1 . in a neighbourhood of the point @xmath260 , according to lemma [ b ] 3 . , the zeta function of the monodromy of @xmath401e^{g\circ\pi})_p$ ] is equal to @xmath270 .
2 . in a neighbourhood of a point @xmath433 , @xmath383 , according to lemma [ c ] , the zeta function of the monodromy of @xmath401e^{g\circ\pi})_{q_\ell}$ ] is equal to @xmath434 .
3 . in a neighbourhood of a point @xmath435 ,
the zeta function of the monodromy of @xmath401e^{g\circ\pi})_q$ ] is equal to @xmath261 .
then , using the mayer - vietoris theorem , we deduce that the zeta function of the monodromy of @xmath419e^{g\circ\pi}))$ ] is equal to @xmath436
let @xmath106 be a formal holonomic @xmath1\langle\partial_t\rangle$]-module .
[ realisation ] there exist a ramification @xmath107 and a regular holonomic @xmath437-module @xmath7 such that @xmath109 is isomorphic to the formalization of @xmath438 .
we recall that we identify @xmath439 with @xmath165 , by choosing a coordinate @xmath26 on @xmath21 .
we begin by stating two lemmas .
[ bb ] 1 .
2 . @xmath441e^{p_2})=\mathcal{o}_{d}[t^{-1}]$ ] .
3 . let @xmath442)$ ] .
let @xmath64 , defined by @xmath65 .
we denote by @xmath5 the curve in @xmath314 parametrized by @xmath443 and by @xmath444 the union of @xmath5 , @xmath33 and @xmath34 .
@xmath445e^{p_2})=\underset{\xi^p=1}{\oplus}\mathcal{o}_{d^{'}}[\tau^{-1}]e^{\alpha(\xi\tau)}\oplus \mathcal{o}_{d^{'}}[\tau^{-1}].\ ] ] we just prove the third point .
the first two points can be proved similary .
let @xmath193 and @xmath194 be the canonical projections and @xmath446 .
first note that @xmath447e^{p_2})=p^{'}_{1+}(\mathcal{o}_{d^{'}\times\mathbb{p}^1}[\ast \rho^{'-1}(\widetilde{z})]e^{p_2^{'}})$ ] ( cf .
lemma [ comm ] ) .
denote by @xmath448e^{p_2^{'}})$ ] the relative de rham complex of @xmath199 concentrated in negative degrees .
it consists in the complex @xmath449\overset{d}{\to}\mathcal{o}_{d^{'}\times\mathbb{p}^1}[\ast \rho^{'-1}(\widetilde{z})],\ ] ] with @xmath450 . by definition ,
@xmath451e^{p_2^{'}})=\mathbb{r}p_{1\ast}^{'}(\mathcal{k}^{\bullet})$ ] .
we consider the spectral sequence @xmath452 , where @xmath453 denotes the complex @xmath454)\overset{d}{\to}r^kp_{1\ast}^{'}(\mathcal{o}_{d^{'}\times\mathbb{p}^1}[\ast \rho^{'-1}(\widetilde{z})])$ ] .
we claim that @xmath454)=0 $ ] , for @xmath455 .
indeed , let @xmath456 and @xmath457 : @xmath458)_p=\\ & = r^kp_{1\ast}^{'}rj_{\ast}j^{-1}(\mathcal{o}_{d^{'}\times\mathbb{p}^1})_p,\\ & = r^k(p_1^{'}\circ j)_{\ast}j^{-1}(\mathcal{o}_{d^{'}\times\mathbb{p}^1})_p,\\ & = \varinjlim_{u\ni p open}r\gamma(j^{-1}(u\times\mathbb{p}^1),j^{-1}(\mathcal{o}_{d^{'}\times\mathbb{p}^1})).\\ \end{array}\ ] ] as @xmath459 is stein , we conclude that @xmath454)=0 $ ] , for @xmath455 .
now @xmath460 is the complex : @xmath461\bigg[\frac{1}{\tau\underset{\xi^p=1}{\prod}(y-\alpha(\xi\tau))}\bigg]\overset{d}{\to } \mathcal{o}_{d^{'}}[y]\bigg[\frac{1}{\tau\underset{\xi^p=1}{\prod}(y-\alpha(\xi\tau))}\bigg],\ ] ] with @xmath462 .
* computation of @xmath463 : + let @xmath464\bigg[\frac{1}{\tau\underset{\xi^p=1}{\prod}(y-\alpha(\xi\tau))}\bigg]$ ] such that @xmath465 .
thus @xmath466 , with @xmath467 $ ] and then @xmath155 .
thus , @xmath468 . *
computation of @xmath469 : + we note first that @xmath470 is generated over @xmath471 $ ] by + @xmath472 .
+ moreover , if @xmath473 , we have : @xmath474 and @xmath475 .
+ then @xmath470 is generated over @xmath471 $ ] by @xmath476 .
but we have @xmath477 .
then , @xmath470 is generated over @xmath471 $ ] by @xmath478 .
+ then we have @xmath479.\frac{1}{(y-\alpha(\xi\tau))}\oplus \mathcal{o}_{d^{'}}[\tau^{-1}]$ ] . + as @xmath480 , we have an isomorphism of @xmath201-modules @xmath481.\frac{1}{(y-\alpha(\xi\tau))}=\mathcal{o}_{d^{'}}[\tau^{-1}]e^{\alpha(\xi\tau)}.\ ] ] + lemma [ bb ] follows .
[ aa ] let @xmath29 be a holonomic @xmath482-module and @xmath7 be a holonomic @xmath437-module .
suppose that @xmath7 is @xmath483-flat .
we have an isomorphism : @xmath484 first , we remark that @xmath485 , with the structure of @xmath437-module given by @xmath486 , where @xmath487 is a local section of the sheaf of vector fields on @xmath314 .
denote by @xmath197 ( resp .
@xmath488 ) the relative de rham functor of @xmath199 ( resp .
the complex of relative forms of @xmath199 ) .
the differential @xmath489 of the complex @xmath490 is defined by : @xmath491 and the action of @xmath492 is given by : @xmath493 then , we have an isomorphism of @xmath492-modules : @xmath494 we conclude with : @xmath495 according to the structure of a formal @xmath1\langle\partial_t\rangle$]-module ( cf .
@xcite ) , we are led to prove the theorem in the cases of regular modules and purely irregular modules . 1 .
@xmath496 suppose @xmath106 regular .
there exist a small disc @xmath21 in @xmath15 centered at the origin and a regular holonomic @xmath497-module @xmath29 , such that the formalization of @xmath166 is @xmath106 ( cf .
theorem 5.3 p. 38 in @xcite ) .
let @xmath7 be the regular holonomic @xmath437-module @xmath498 .
we have : @xmath499 2 .
@xmath500 let @xmath501)$ ] and @xmath502 be a regular holonomic @xmath503\langle\partial_{\tau}\rangle$]-module , such that @xmath504=\widehat{r_{\alpha}}$ ] .
let @xmath107 and suppose @xmath505 .
+ there exist a small disc @xmath506 in @xmath15 centered at the origin and a regular holonomic @xmath201-module @xmath507 such that the formalization of @xmath508 is @xmath502 ( cf .
theorem 5.3 p. 38
in @xcite).but there exists a regular holonomic @xmath482-module @xmath79 such that @xmath509 , where @xmath64 , @xmath65 . indeed ,
as @xmath510 is exact and @xmath507 is regular holonomic with @xmath511=\widetilde{r_{\alpha}}$ ] , we can reduce the proof to the case where @xmath512 , with @xmath513 and @xmath514 ( cf .
corollary 19 in @xcite ) .
then , we consider @xmath515 .
+ using notations of lemma [ bb ] , we denote by @xmath7 the regular holonomic @xmath437-module @xmath516}/{p_1^{\ast}(r_{\alpha})[t^{-1}]}}$ ] . according to lemmas [ aa ] and [ bb ] ,
we have two isomorphisms : @xmath517e^{p_2})=\underset{\xi^p=1}{\oplus}\widetilde{r_{\alpha}}e^{\alpha(\xi\tau)}\oplus \widetilde{r_{\alpha}}\\ \rho^{\ast}p_{1+}(p_1^{\ast}(r_{\alpha})[t^{-1}]e^{p_2})=\widetilde{r_{\alpha}}. \end{array}\ ] ] + then the formalization of @xmath518 is @xmath519 .
in this section , we prove theorem [ 2 ] . we consider
@xmath520 , the extension of @xmath521 at infinity , where @xmath522 .
let @xmath523 .
the proof will be divided in two steps . 1 .
we consider the following diagrams @xmath524 ^ -i\ar[d]^-{\pi_1}&\mathbb{p}^1\times\mathbb{p}^1\ar[d]^-{p_1}&b_c\times\mathbb{p}^1\ar@{_(->}[l]_-{i_c}\ar[d]^{p_1 } & \mathbb{c}^2\ar@{^(->}[r]^-i\ar[d]^-{\pi_2}&\mathbb{p}^1\times\mathbb{p}^1\ar[d]^-{p_2}&b_c\times\mathbb{p}^1\ar@{_(->}[l]_-{i_c}\ar[d]^{p_2}\\ \mathbb{c}\ar@{^(->}[r]^-j&\mathbb{p}^1&b_c\ar@{_(->}[l]_-{j_c},&\mathbb{c}\ar@{^(->}[r]^-j&\mathbb{p}^1&\mathbb{p}^1\ar@{=}[l],}\ ] ] where @xmath432 , @xmath525 ( resp .
@xmath199 , @xmath526 ) are the two canonical projections .
+ then we have : @xmath527 + as the formalization of the germ @xmath528 and of its analytization are equal , @xmath528 and @xmath529 have the same formal irregular part .
2 . then we claim that @xmath529 and @xmath123 have the same formal irregular part . according to the formal decomposition theorem ( cf .
theorem 1.2 p. 43 and theorem 2.3 p. 51 in @xcite ) , there exists a ramification @xmath530 , @xmath531 , such that the modules @xmath532 and @xmath533 are isomorphic to the finite direct sum of some formal modules of exponential type .
then , it is sufficient to prove that , given an element @xmath534 $ ] , the regular part of @xmath535 and @xmath536 are isomorphic .
thus , we are led to prove that the specialization at @xmath126 of @xmath537 and @xmath538 are isomorphic ( cf .
example 5.2.1 in @xcite ) .
+ let @xmath539 .
we denote by @xmath540 and @xmath541 the canonical projection .
let @xmath542 .
we have : @xmath543),\\ & \text{(p.5 in \cite{ma3 } ) . }
\end{array}\ ] ] + according to proposition [ 11 ] extended to complexes of @xmath0-modules , the support of @xmath544 is included in @xmath132 .
then , + @xmath545)_{(c,\infty)},\\ & = ( \operatorname{dr}~\mathcal{h}^k\operatorname{sp}_{\{c\}\times\mathbb{p}^1}(\rho^{'\ast}(\mathcal{p}_c)e^{p_2^{'}-\alpha\circ p_1^{'}})[+1])_{(c,\infty)},\\ & = r^0\gamma(\{c\}\times\mathbb{p}^1,\operatorname{dr}~\mathcal{h}^k\operatorname{sp}_{\{c\}\times\mathbb{p}^1}(\rho^{'\ast}(\mathcal{p}_c)e^{p_2^{'}-\alpha\circ p_1^{'}})[+1]),\\ & = r^0\gamma(\{c\}\times\mathbb{p}^1,\operatorname{dr}~\operatorname{sp}_{\{c\}\times\mathbb{p}^1}(\rho^{'\ast}(\mathcal{h}^k\mathcal{p}_c)e^{p_2^{'}-\alpha\circ p_1^{'}})[+1]),\\ & = \mathcal{h}^0\overline{p}_{1+}^{'}\operatorname{sp}_{\{c\}\times\mathbb{p}^1}(\rho^{'\ast}(\mathcal{h}^k\mathcal{p}_c)e^{p_2^{'}-\alpha\circ p_1^{'}})_c,\\ & \text{(p.5 in \cite{ma3}),}\\ & = \operatorname{sp}_c(\mathcal{h}^0p_{1+}^{'}(\rho^{'\ast}(\mathcal{h}^k\mathcal{p}_c)e^{p_2^{'}-\alpha\circ p_1^{'}}))_c,\\ & \text{(theorem 4.8.1 p. 226 in \cite{mesa}),}\\ &
= \operatorname{sp}_c(\mathcal{h}^0p_{1+}^{'}(\rho^{'\ast}(\mathcal{h}^k\mathcal{p}_c)e^{p_2^{'}})e^{-\alpha})_c,\\ & = \operatorname{sp}_c((\rho^{\ast}(\mathcal{h}^0p_{1+}(\mathcal{h}^k\mathcal{p}_ce^{p_2})))e^{-\alpha})_c,\\ & \text{(lemma \ref{comm } ) . }
\end{array}\ ] ] 3 borel , a. : algebraic d - modules , perspectives in math * 2 * , academic press , boston ( 1987 ) .
brianon , j. , maisonobe , ph . : idaux de germes doprateurs diffrentiels une variable , ens . math . * 30 * , p. 7
- 38 ( 1984 ) .
kashiwara , m. : on the maximally overdetermined systems of differential equations , publ .
rims * 10 * , kyoto univ .
, p. 563 - 579
kashiwara , m. : vanishing cycle sheaves and holonomic systems of differential equations . in : algebraic geometry , tokyo / kyoto , 1982 , p. 134 - 142 ,
lecture notes in math . * 1016 * , springer , berlin ( 1983 ) .
kashiwara , m. : index theorem for constructible sheaves . in : differential systems and singularities ,
luminy , 1983 , p. 193 - 209 , astrisque * 130 * , soc .
france , paris ( 1985 ) .
kashiwara , m. , schapira , p. : sheaves on manifolds , grundlehren der mathematischen wissenschaften * 292 * , springer - verlag ( 1990 ) .
laurent , y. , malgrange , b. : cycles proches , spcialisation et @xmath0-modules .
fourier * t. 45 , n. 5 * , p. 1353 - 140
maisonobe , ph.,mebkhout , z. : le thorme de comparaison pour les cycles vanescents . in : lments de la thorie des systmes diffrentiels gomriques , p. 311 - 389 , sminaires et congrs * 8 * , soc .
france , paris ( 2004 ) .
malgrange , b. : polynme de bernstein - sato et cohomologie vanescente . in analysis and topology on singular spaces , ii , iii , luminy , 1981 , p. 243 - 267 , astrisque * 101 * , soc . math .
france , paris ( 1983 ) .
malgrange , b. : equations diffrentielles coefficients polynomiaux .
progress in math . * 96 * , birkhuser , boston ( 1991 ) .
malgrange , b. : de rham complex and direct images of @xmath0-modules . in images
directes et constructibilt , p. 1 - 13 , travaux en cours * 46 * , hermann , paris , ( 1993 ) .
mebkhout , z. : le formalisme des six oprations de grothendieck pour les @xmath4-modules cohrents .
travaux en cours * 35 * , hermann , paris ( 1988 ) .
mebkhout , z. , narvez - macarro , l. : le thorme de constructibilit de kashiwara . in : image
directe et constructibilit , p. 47 - 98 , travaux en cours * 46 * , hermann , paris ( 1993 ) .
roucairol , c. : lirrgularit du complexe @xmath546 , thesis * n. 619 * , angers university ( 2004 ) roucairol , c. : irregularity of an analogue of the gauss - manin systems , bull .
france * 134 ( 2 ) * , p. 269 - 286
roucairol , c. : the irregularity of the direct image of some @xmath0-modules , publ .
rims * 42 ( 4 ) * , kyoto univ .
sabbah , c. : @xmath4-modules et cycles vanescents . in @xcite , p. 201 - 239
sabbah , c. : algebraic theory of differential equations . in @xmath0-modules cohrents et holonomes , p. 1 - 80 , les cours du cimpa * 45 * , hermann , paris ( 1993 ) .
sabbah , c. : equations diffrentielles points singuliers irrguliers et phnomne de stokes en dimension @xmath547 .
astrisque * 263 * , soc .
france ( 2000 ) . | we compute formal invariants associated with the cohomology sheaves of the direct image of holonomic @xmath0-modules of exponential type .
we also prove that every formal @xmath1\langle\partial_t\rangle$]-modules is isomorphic , after a ramification , to a germ of formalized direct image of analytic @xmath0-module of exponential type . |
it is widely recognized that the ferromagnetic ising chain with first - neighbor interactions and glauber dynamics is one of the simplest , exactly soluble , stochastic dynamical systems @xcite . at finite temperatures , in the stationary regime , the two - time spin autocorrelation , @xmath3 , and the associated response function , @xmath4 , of this glauber chain are time - translationally invariant , and duly related by the usual expression of the fluctuation - dissipation theorem @xcite .
there is also an aging regime , with violation of the usual form of the fluctuation - dissipation theorem . at the critical point , at zero temperature , for large values of the observation time @xmath5 , the fluctuation - dissipation ratio of the glauber chain assumes the non trivial limoting value @xmath6 .
we now revisit a linearized version of the glauber model , proposed by one of us a few years ago @xcite . as in the original glauber model on a @xmath7-dimensional hypercubic lattice , we still consider a one - flip stochastic process .
each site @xmath8 of the lattice is associated with a spin variable @xmath9 . however
, the time evolution is now governed by a linear spin - flip ratio , @xmath10 , \label{rate}\ ] ] where @xmath11 $ ] is a parameter , the sum is over the @xmath12 nearest neighbors of site @xmath13 , the time scale is set by the parameter @xmath14 , and @xmath15 stands for a configuration of spin variables , @xmath16 .
the evolution of the probability @xmath17 of the spin configuration @xmath15 at time @xmath5 is given by the master equation @xmath18 , \label{master}\ ] ] where @xmath19 is defined as the configuration @xmath15 with @xmath20 replaced by @xmath21 . from these equations , it is straightforward to calculate exact analytical expression for the site magnetization , @xmath22 , and the pair correlation function , @xmath23 .
in contrast to the original glauber model , defined by a non linear transition rate , with exact solutions restricted to @xmath24 , we can now write expressions for @xmath25 and @xmath26 in all dimensions @xmath0 . for @xmath27 ,
the linearized model displays a disordered ( paramagnetic ) phase , with exponentially decaying pair correlations ; at the critical point @xmath28 , correlations decay algebraically @xcite .
it is important to distinguish between this linear glauber model and the so - called glauber dynamics , associated with a nonlinear transition rate , and which is used to simulate the ising model .
the linear glauber model may be regarded as a voter model with noise @xcite .
it displays just one phase , for all dimensions , as long as noise is finite . in the absence of noise ( @xmath28 ) it becomes critical . in one dimension
there is a close identification between these two versions of the glauber model . for @xmath1 , however , in contrast to the original model , the analytically solvable linear glauber model is microscopically irreversible ( in other words , although having a stationary state , it does not obey detailed balance , and can not be associated with a hamiltonian ) .
the fluctuation - dissipation theorem is usually conceived for systems that do obey detailed balance callen51,kubo57 .
it is then reasonable to ask some questions , including the validity of the fluctuation - dissipation theorem and the presence of an aging regime , about the dynamical behavior of systems that do not obey detailed balance .
one of the purposes of this article is to carry out a thorough analytical investigation of a particular system , as the non linear glauber model in @xmath1 dimensions , which belongs to the large class of microscopically irreversible models oliveira03,tome89,tome91,oliveira92,oliveira93 .
the dynamical calculations of interest in this investigation are performed in the presence of a ( small ) perturbation . in the treatment of stochastic models
it is natural to introduce the modified one - spin - flip rate , @xmath29 where @xmath30 is the unperturbed flipping rate associated with the @xmath13th spin , and @xmath31 is the time - dependent disturbance coupled to the dynamic variable @xmath32 . alternatively , as the calculations are restricted to small perturbations
, we can write @xmath33 if the model is microscopically reversible , that is , if the unperturbed transition rate @xmath34 obeys detailed balance , there is a model hamiltonian @xmath35 , and it is straightforward to show that the perturbed transition rate @xmath36 , given by eq .
( [ 1 ] ) , also obeys detailed balance . in this reversible case ,
the model is described by the hamiltonian @xmath37 , where @xmath38 , and is @xmath39 proportional to the inverse of the temperature .
this form of perturbation is then suitable for reversible models , with a disturbance @xmath31 proportional to the external field . assuming this expression for the flipping rate , the fluctuation - dissipation theorem is given by @xmath40 where @xmath41 is the response function
, @xmath25 is the average of @xmath32 at the observation time @xmath5 , and @xmath42 is the autocorrelation function of @xmath32 between the observation time @xmath5 and the waiting time @xmath43 ( with @xmath44 ) . for systems obeying detailed balance , we have @xmath45 , and relation ( [ 2a ] ) reduces to the usual form of the fluctuation - dissipation theorem .
another version of the fluctuation - dissipation theorem relates the susceptibility , associated with a given dynamical variable @xmath46 , such as the total magnetization , and its variance , @xmath47 where @xmath48 is a static , time - independent disturbance , introduced by the prescription of eq . .
in this work , we show that both forms of the fluctuation - dissipation relation , and , are valid for the linear glauber model in the stationary regime . in the transient regime , where aging behavior takes place ,
these relations are no longer obeyed .
it is then appropriate to define @xcite a fluctuation - dissipation ratio , @xmath49 in the linear glauber model , for all values of the dimension @xmath0 , we show that @xmath50 in the limit @xmath51 , except at the critical point , @xmath28 , in which case @xmath52 .
the layout of this paper is as follows . some results for the linear glauber model , including a discussion of the lack of detailed balance for @xmath1 , and calculations of the site magnetization and the spatial two - body correlations , are reviewed is section 2 .
these one - time functions play a major role in the calculations of the two - time functions , as the autocorrelation and the response functions , which are obtained in section 3 . in this section
, we also consider the stationary limit and make a number of comments on the non - stationary regime .
section 4 contains some conclusions .
we have already mentioned that the linear glauber model has been introduced by one of us @xcite as an extension of the original glauber model @xcite , and may be regarded as a voter model with noise .
this linear glauber model is defined by the ( linear ) one - spin - flip rate given by eq .
( [ rate ] ) , which should be inserted into the master equation ( master ) .
> from equations ( [ rate ] ) and ( [ master ] ) , it is not difficult to obtain the evolution equations for the site magnetization and the spatial pair correlation , @xmath53 and @xmath54 , \label{4b}\ ] ] for @xmath55 . if @xmath13 and @xmath56 are nearest - neighbor sites , the right - hand side of eq .
( [ 4b ] ) contains terms like @xmath57 and @xmath58 which should be set equal to @xmath59 .
the possibility of obtaining these exact expressions , and of performing the exact calculations that are going to be reported in this article , is one of the most relevant features of the linear glauber model .
as shown by oliveira @xcite , for @xmath27 this model displays a disordered ( paramagnetic ) phase with exponentially decaying correlations . for @xmath60 , it becomes critical , with algebraically decaying correlations at @xmath28 .
the linear glauber model in one dimension has a reversible dynamics . in one dimension , the probability of occurrence of any sequence of states and the probability of the associated reverse sequence of states are the same . in dimensions larger than one ,
this is no longer valid .
consider , for instance , the four states shown in fig .
[ seq ] , on a square lattice .
suppose that the system follows the sequence of states @xmath61 , @xmath62 , @xmath63 , @xmath64 , and returns to the initial state @xmath61 . if the interval @xmath65 between two successive states is small , then , according to the spin - flip rate given by eq .
( [ 1 ] ) , the probability of occurrence of the sequence a@xmath66b@xmath67c@xmath66d@xmath66a is given by @xmath68 on the other hand , the reversed sequence , a@xmath66d@xmath66c@xmath67b@xmath66a , has the probability @xmath69 these two probabilities are distinct , so that the linear glauber model on a square lattice is indeed irreversible , the only exception being the trivial case @xmath70 . a generalization of this result to larger dimensions
is easily found by filling the sites created by the introduction of more dimensions with `` @xmath71 spins '' .
hence , the detailed balance can not be valid , and the stationary state is _ a priori _ not known .
the connection of the transition rates with a gibbs measure , as it has been possible in the one - dimensional case , is now forbidden .
we now introduce the fourier transform of @xmath25 , @xmath72 and the laplace transform of @xmath73 , @xmath74 using these transforms , the differential equation is reduced to the algebraic form @xmath75 where @xmath76 and @xmath77 is the fourier - laplace transform of @xmath78 .
the inverse laplace transformation leads to @xmath79 inverse fourier transforming , we then have @xmath80 where @xmath81
we can also obtain the site magnetization with reference to an initial time @xmath82 instead of @xmath83 . in this case , we just write @xmath84 where @xmath85 is the site magnetization at time @xmath43 .
if we look for translationally invariant solutions of eq . , the spatial correlation between sites @xmath13 and @xmath56 will be a function of distance , @xmath86 .
we then write eq .
as @xmath87 for @xmath88 , with @xmath89 whenever it appears on the right - hand side . using a method introduced by oliveira @xcite ,
let us write an equation for @xmath90 , @xmath91 where @xmath92 is chosen to ensure that @xmath89 . actually , @xmath92 is defined by @xmath93 consequently , eqs . and can be written as @xmath94 for all values of @xmath13 .
eq . can be examined from two points of view . on the one hand
, it is a first - order differential equations in time . on the other hand
, the summation over nearest neighbors resembles a discrete lattice laplacian , and from this point of view it is a discrete second - order difference equation that can be solved by the use of a green function .
we then introduce the laplace transform of @xmath95 , @xmath96 so that eq . may be written as @xmath97 = -2\hat{q}_{r}(s)+% \frac{\lambda } { d}\sum_{\delta } \hat{q}_{r+\delta } ( s)+\hat{b}(s)\delta _ { r,0 } , \label{11x}\ ] ] where @xmath98 is chosen such that @xmath89 , which is equivalent to take @xmath99 . for a random initial condition , which corresponds to a quench from infinite temperature ,
it is appropriate to take @xmath100 . to solve equation
, we introduce the lattice green function associated with a @xmath0-dimensional hypercubic lattice , @xmath101 where the integration is over the first brillouin zone , and @xmath102 this green function satisfies the equation @xmath103 in terms of this lattice green function , the solution of eq .
can be written as @xmath104 where @xmath98 should be chosen so that @xmath99 .
this leads to @xmath105 , \label{16x}\ ] ] from which it follows @xcite the solution @xmath106 + \frac{\hat{g}_{r}(s)}{s\hat{g}_{0}(s)}. \label{17a}\ ] ] for a completely random initial condition , @xmath107 , we have @xmath108 , @xmath109 and @xmath110
the calculation of the two - time response function , @xmath111 requires the application of a small perturbation , which is introduced according to the prescription , and from which one measures the response of the system . as pointed out in the introduction , we assume a perturbed spin - flip rate , given by @xmath10 \left [ 1-h_{r}(t)\sigma _ { r}% \right ] , \label{25}\ ] ] where @xmath112 is a time - dependent disturbance coupled to the dynamic variable @xmath113 .
we then write the equation of motion for the site magnetization , @xmath114 , \label{26}\ ] ] which can also be written as @xmath115 where @xmath92 is given by eq . .
using the same procedures adopted to solve eq . , and taking into account that the last term in eq .
is a known function of time , we have @xmath116 > from this expression , we calculate @xmath117 which leads to the response function , @xmath118 the correlation @xmath119 of a spin at a given site @xmath13 , at time @xmath43 , with the same spin at a later time @xmath5 ( @xmath44 ) is formally written as @xmath120 where @xmath121 is the conditional probability of finding the configuration @xmath15 at time @xmath5 given the configuration @xmath122 at an earlier time @xmath43 .
noting that the site magnetization , @xmath123 , with the initial condition @xmath124 , may be written as @xmath125 and using eq .
, we have @xmath126 which can be inserted into eq . to give @xmath127 which finally leads to @xmath128 in the stationary regime the waiting time @xmath43 and the observation time @xmath5 grow without limits , but the difference @xmath129 is fixed . to be more precise , @xmath130 , with @xmath131 and @xmath132 fixed .
> from the laplace transform final value theorem , we have @xmath133 so that @xmath134 which can be written as @xmath135 on the other hand , taking into account that @xmath136 the response function can be written as @xmath137 using the definition of @xmath138 , given by , we have @xmath139 both quantities , @xmath140 and @xmath141 , are time - translationally invariant ( functions of @xmath142 only ) , as it should be anticipated in a stationary regime . moreover , the fluctuation - dissipation theorem is trivially satisfied , with @xmath143 . for large time differences , and @xmath144 ,
it is easy to see that both the autocorrelation and the response functions decay exponentially , according to @xmath145 $ ] , with the equilibration time @xmath146 $ ] .
the fluctuation - dissipation theorem can also be written in terms of global variables , as the magnetization and the corresponding answer to a static perturbation .
let @xmath46 denote the magnetization , @xmath147 and let us consider a static homogeneous disturbance @xmath48 , defined by eq . .
then @xmath148 where @xmath149 is a variance , @xmath150 .\ ] ] due to the translational invariance of the lattice , @xmath151 .
this quantity is the solution of eq .
, which is now given by @xmath152 with the stationary solution @xmath153 for @xmath154 , the magnetization vanishes as @xmath155 , so that the variance is given by @xmath156 > from eq .
it follows that @xmath157 with the stationary solution @xmath158 > from equations and , it is seen that relation is clearly satisfied in the stationary regime .
the temporal behavior of the autocorrelation and the response functions , as calculated in this section , already suggests the existence of an aging regime .
equations ( [ 29 ] ) and ( [ 37 ] ) , for @xmath159 and @xmath160 , are valid for all values of @xmath43 and @xmath5 , with @xmath161 .
the dependence of these functions on both @xmath5 and @xmath43 , and not on @xmath142 only , leads to the existence of aging . in this regime ,
the role of the spatial correlations is still crucial , since they are responsible for the realization of the aging scenario . in the stationary regime
, it should be noted that @xmath162 becomes a time - independent quantity in the @xmath130 limit only , and this is the reason of the dependence of the two - time functions on @xmath163 only . in the transient regime , we do not expect the validity of the usual form of the fluctuation - dissipation relation given by eq . .
it has been convenient @xcite to characterize the distance to the stationary regime by the fluctuation - dissipation ratio , @xmath49 a particular interesting quantity is the limit @xmath164 where @xmath165 > from the results of this section , it is easy to write @xmath166 where @xmath167 is the sum given by . at the critical point , @xmath28 , it follows that @xmath168 for any time @xmath43 , so that @xmath2 . in the disordered phase , @xmath169 , eqs . and may be used to conclude that @xmath170 .
we have reported a number of exact calculations for the dynamical behavior of a @xmath0-dimensional linearized version of the stochastic glauber model . in one dimension ,
both the linear and the original model are essentially equivalent . for @xmath1
, however , the rates of transition of the linear glauber model do not obey the conditions of detailed balance .
this linear model can be regarded as a voter model with noise ; it displays just one stable phase as long as noise is finite , and becomes critical in the absence of noise ( @xmath28 ) . since the dynamical properties are usually conceived for systems that do obey detailed balance , we decided that it was appropriate to carry out some explicit calculations for a microscopically irreversible model .
we have obtained expressions for the two - time autocorrelation , @xmath171 , and response functions , @xmath172 , which depend on both observation time @xmath5 and waiting time @xmath173 . in the stationary , infinite time limit , @xmath174 , in the presence of noise ( @xmath175 ) , the spatial correlations are independent of time , and the two - time functions become translationally invariant ( depending on the difference @xmath176 only ) .
the usual form of the fluctuation - dissipation theorem , @xmath177 , is trivially observed in this regime . in the scaling regime ( @xmath178 ) , for @xmath27 , we obtain a non trivial fluctuation - dissipation ratio , @xmath179 . at the critical point , @xmath28
, we have @xmath180 , which further indicates that the dynamical behavior of the @xmath0-dimensional microscopically irreversible linear glauber model is very similar to the behavior of its one - dimensional reversible counterpart . | we obtain exact expressions for the two - time autocorrelation and response functions of the @xmath0-dimensional linear glauber model .
although this linear model does not obey detailed balance in dimensions @xmath1 , we show that the usual form of the fluctuation - dissipation ratio still holds in the stationary regime . in the transient regime
, we show the occurence of aging , with a special limit of the fluctuation - dissipation ratio , @xmath2 , for a quench at the critical point .
pacs numbers : 02.50.ey , 05.50.+q , 05.70.ln , 75.10.hk emails : [email protected] |
0.2 cm ( i ) -1.0 cm 0.5 cm + 0.2 cm ( ii ) -0.3 cm 1.6 cm -1.3 cm the pibeta experiment at the paul scherrer institute ( psi ) is a collaboration of seven institutions that collected the world largest sample of rare pion and muon decays during the 1999 - 2001 and 2004 beam periods @xcite .
the pibeta detector system is based on a large acceptance 240-module pure csi electromagnetic shower calorimeter .
the detector includes with an active degrader ad , a segmented active target at , a 20-bar cylindrical plastic scintillator veto pv for particle identification , a pair of tracking cylindrical multi - wire proportional chambers mwpc1/2 and an active cosmic veto shield @xcite . a schematic drawing of the detector is shown in fig . [ fig : det ] .
the incident 114mev / c @xmath6 beam with a minimal @xmath7/@xmath8 contamination was tagged with a thin forward beam counter bc , slowed down in the degrader and ultimately stopped in a tight @xmath9 mm spot in the active target .
the 1999 - 2001 runs , optimized for the pion beta decay measurement , used beam fluxes of up to 1mhz .
beam intensities of 50 - 200@xmath6/s were used in the 2004 for optimal acquisition of radiative decay events .
the recorded data , comprising @xmath10 @xmath6 stops , were obtained by a dedicated two - arm high - threshold trigger as well as 11 physics and calibration triggers , some of which were prescaled .
events.,title="fig:",width=283 ] + -0.8 cm events.,title="fig:",width=283 ] 0.7 cm pion beta decay ( @xmath0 ) rate offers one of the most precise means of testing the conserved vector current hypothesis ( cvc ) @xcite and studying the weak @xmath11-@xmath12 quark mixing @xcite .
the standard model ( sm ) description of the @xmath13 decay is theoretically unambiguous within a 0.1% uncertainty @xcite , but a small @xmath14 branching ratio poses significant experimental chalenges .
the 3.9% uncertainty of the previous most precise measurement , made using the @xmath15 spectrometer at lampf @xcite , was not accurate enough to test the full extent of radiative corrections which stand at @xmath16% @xcite .
the fast analog hardware triggers were designed to accept nearly all non - prompt @xmath13 and and a sample of prescaled @xmath17 events with individual shower energy exceeding the michel endpoint ( high threshold @xmath18 52 mev ) .
the data analysis provided clean distributions of 64,047 @xmath13 decay events which agreed very well with energy , angular and timing spectra predicted by the geant3 monte carlo detector simulations .
the cosmic muon , prompt , radiative pion and accidental backgrounds were determined to be @xmath19 of the @xmath13 signal , fig .
[ fig : pb ] .
we have chosen to normalize the @xmath13 yield to the yield of @xmath17 events whose branching ratio is known with 0.33% uncertainty experimentally @xcite and @xmath20% accuracy theoretically @xcite . using the pdg @xcite recommended value of @xmath21
, we find the pion beta branching ratio @xcite : @xmath22 \cdot 10^{-8}.\ ] ] when normalizing to the theoretical value @xmath23@xcite we obtain : @xmath24 \cdot 10^{-8}.\ ] ] our result for @xmath25 is in excellent agreement with the prediction of the sm : @xmath26 and stands as the most accurate confirmation of the cvc in a meson to date .
+ -0.6 cm + -0.6 cm -0.5 cm [ cols= " < , < , < " , ] + the fitted @xmath27 values are based on 2004 data set .
the radiative muon decay @xmath4 measurement provides another critical consistency check of overall analysis . in the standard model
this process is parameterized via michel parameters all of which , save @xmath28 , can be determined from the ordinary muon decay @xcite .
a non - zero value of @xmath28 would imply the non-(@xmath29 ) structure of the electroweak interaction .
the most recent direct measurement of @xmath28 can be interpreted as an upper limit of 0.141 ( at 90% cl ) @xcite .
our two - dimensional michel parameter fit ( @xmath30 ) of the @xmath31 radiative muon events collected in 2004 ( fig .
[ fig : radmu ] ) corresponds to the upper limit @xmath32 and simultaneously yields the sm value @xmath33 @xcite .
the details are summarized in table [ tab2 ] .
events in csi calorimeter ( top ) and timing distribution fit to @xmath17 candidate events ( bottom).,title="fig:",width=283 ] + events in csi calorimeter ( top ) and timing distribution fit to @xmath17 candidate events ( bottom).,title="fig:",width=283 ] -1.0 cm we have proposed to perform a new precise measurement of the @xmath17 branching ratio at psi using a suitably upgraded pibeta detector system . the experiment has been approved with high priority in 2006 .
the measurement is motivated by the fact that at present accuracy of that branching ratio lags behind the theoretical precision by an order of magnitude .
we will build on our past experience in exploiting the @xmath34 decay for normalizing purposes ( fig .
[ fig : p2e ] ) .
we plan to rebuild the target region of the detector and develop the new digitizing detector for the beam counters .
the proposed accuracy of the new measurement is @xmath35 or lower .
we have reported new and improved absolute branching ratios for the following rare decays : ( 1 ) @xmath0 , ( 2 ) @xmath1 , and ( 3 ) @xmath4 .
the yields of @xmath17 and @xmath36 decays that were used for normalization are also internally consistent when compared to the total measured number of decaying @xmath6 s and @xmath8 s @xcite .
our results confirm the cvc hypothesis in the @xmath6 system at 0.55% level , rule out the tensor contribution in the radiative @xmath6 decay with the form factor @xmath37 ( 90% cl ) , and set the new 90% cl limit on the parameter @xmath32 in the radiative @xmath8 decay .
9 pibeta home page : http://pibeta.phys.virginia.edu .
e. frle , d. poani , k.a .
assamagan et al . , nucl .
inst . and meth . a 526 ( 2004 ) 300 .
r. p. feynman and m. gell - mann , phys .
109 ( 1958 ) 193 .
n. cabibbo , phys .
( 1963 ) 531 .
w. jaus , phys .
d 63 ( 2001 ) 053009 . v. cirigliano , m. knecht , h. neufeld , and h. pichl , eur .
j. c 27 ( 2003 ) 255 .
mcfarlane , l.b .
auerbach , f.c .
gaille et al . , phys .
d 32 ( 1985 ) 547 . w. j. marciano and a. sirlin , phys .
( 1986 ) 22 .
britton , s. ahmad , d.a .
bryman at al .
68 ( 1992 ) 3000 .
g. czapek , a. federspiel , a. flkiger et al .
70 ( 1993 ) 17 .
marciano , a. sirlin , phys .
71 ( 1993 ) 3629 .
r. decker and m. finkemeier , nucl .
b 438 ( 1995 ) 17 .
s. eidelman et al .
, physics letters b592 ( 2004 ) 1 .
d. poani , e. frle , v.a .
baranov et al . , phys .
93 ( 2004 ) 181803 - 1 .
e. frle , d. poani , v.a .
baranov et al . , phys .
93 ( 2004 ) 181804 - 1 .
e. a. kuraev , yu . m. bystritsky and e. p. velicheva , private communication ( 2003 ) .
v. n. bolotov , s. n. gninenko , r. m. djilkibaev et al .
b243 ( 1990 ) 308 .
a. a. poblaguev , phys .
b238 ( 1990 ) 108 .
a. a. poblaguev , phys .
b286 ( 1992 ) 169 .
bychkov , ph .
d. thesis , university of virginia ( 2005 ) .
poblaguev , phys .
d 68 ( 2003 ) 054020 . k. mursula and f. schenk , nucl .
b253 ( 1985 ) 189 . w. eichenberger , r. engfer , and a. van der schaaf , nucl .
a , 412 ( 1984 ) 523 .
vandevender , ph .
d. thesis , university of virginia ( 2006 ) .
e. frle , fizika b 13 ( 2004 ) 243 . | the pibeta collaboration has used a non - magnetic pure csi calorimeter operating at the paul scherrer institute to collect the world s largest sample of rare pion and muon decays .
we have extracted the absolute @xmath0 decay branching ratio with the 0.55% total uncertainty .
the @xmath1 data set was used to extract weak axial and vector form factors @xmath2 and @xmath3 , yielding a significant improvement in the precision of @xmath2 and @xmath3 .
the @xmath4 distributions were well described with the two - parameter @xmath5 ) solution .
these results bring major improvements in accuracy over the current particle data group listings and agree well with the predictions of the standard model . |
minimal surfaces have recently become subject of intensive study in physical and biological sciences , e.g. materials science and molecular engineering which is due to its area minimizing property .
they are used in modeling physical phenomena as soap films , block copolymers , protein folding , solar cells , nanoporous membranes , etc .
minimal surfaces found applications also in architecture , cagd , and computer graphics where bzierpolynomials and splines are widely used to efficiently describing , representing and visualizing 3d objects .
hence it is important to know minimal surfaces in polynomial form of lower degrees .
bi - cubic polynomial minimal surfaces are studied in @xcite .
polynomial surfaces of degree 5 and 6 are studied in @xcite and @xcite where some interesting surfaces are described and their properties are examined .
examples of polynomial minimal surfaces of arbitrary degree are presented in @xcite . in section 3
we specify the result of cosn and monterde @xcite concerning bi - cubic polynomial minimal surfaces .
we note that their proof that these surfaces coincide up to affine transformation with the classical enneper surface concerns the case of surfaces in _ isothermal _ parameters . in section 4
we consider an analogous problem for polynomial surfaces defined by charts @xmath0 of degree 4 on both @xmath1 and @xmath2 .
it turns out that they are more various than the bi - cubic ones , so they may be more useful in computer graphics . in this paper
we use a new approach to minimal surfaces proposed by ganchev @xcite as well as the method from @xcite to obtain a parametrization of the surface in canonical parameters .
in section 5 we consider bi - quartic harmonic bzier surfaces .
we show that for a special choice of nine boundary control points the corresponding harmonic bzier surface is uniquely determined and is symmetric with respect to one of the coordinate planes @xmath3 ,
@xmath4 , and @xmath5 . based on the bzier representation we apply computer modeling and visualization tools to illustrate and support our results .
let @xmath6 be a regular surface
. then @xmath6 is locally defined by a chart @xmath7 as usual we denote by @xmath8 , @xmath9 , @xmath10 , ... the partial derivatives of the vector function @xmath11 .
then the coefficients of the first fundamental form are given by the inner products @xmath12 and the unit normal is @xmath13 then the coefficients of the second fundamental form are defined by @xmath14 the gauss curvature @xmath15 and the mean curvature @xmath16 of @xmath6 are given by @xmath17 respectively . note that the gauss curvature and the mean curvature of a surface do not depend on the chart .
the surface @xmath6 is said to be _ minimal _ if its mean curvature vanishes identically . in this case
the gauss curvature is negative and the normal curvature of @xmath6 is the function @xmath18 , see @xcite .
we say that the chart @xmath0 is _ isothermal _ or that the parameters @xmath19 are isothermal if @xmath20 , @xmath21 .
it is always possible to change the parameters @xmath19 so that the resulting chart be isothermal .
we note however that this change of the parameters to isothermal ones is in general nonlinear . when the chart is isothermal it is possible to use complex functions to investigate it .
we shall explain briefly this .
namely let @xmath22 and @xmath23 be two holomorphic functions ( actually sometimes they are taken meromorphic ) .
define the weierstrass complex curve @xmath24 by @xmath25 then @xmath24 is a minimal curve , i.e. @xmath26 , and its real and imaginary parts @xmath27 are minimal charts .
moreover , they are isothermal and are harmonic functions ( i.e. @xmath28 , @xmath29 , where @xmath30 is the laplace operator ) as the real and complex part of a holomorphic function .
conversely , every minimal surface can be defined at least locally in this way .
of course a minimal surface can be generated by the weierstrass formula with different pairs of complex functions @xmath22 , @xmath23 .
it is easy to see that the coefficients of the first fundamental form of a chart defined via the weierstrass formula with functions @xmath22 , @xmath23 are given by @xmath31 the normal curvature is computed to be @xmath32 see @xcite , theorem 22.33 .
recently ganchev @xcite has proposed a new approach to minimal surfaces . briefly speaking he introduces special parameters called _ canonical principal parameters_.
a chart with such parameters is isothermal .
moreover , the coefficients of the two fundamental forms are given by @xmath33 l=1 , & \
m=0 , & \ n=-1 .
\end{array}\ ] ] his idea leads to the fact that the real part of the minimal curve @xmath34 is a minimal surface in canonical principal parameters . note that this is the weierstrass formula with @xmath35 , @xmath36 .
we shall use also the following theorems : * theorem a. * @xcite if a surface is parametrized with canonical principal parameters , then the normal curvature satisfies the equation @xmath37 conversely , for any solution @xmath38 of equation there exists a * unique * ( up to position in the space ) minimal surface with normal curvature @xmath38 , where @xmath19 are canonical principal parameters.@xmath39 * theorem b. * @xcite let the minimal surface @xmath6 be defined by the real part of .
any solution of the differential equation @xmath40 defines a change of the isothermal parameters of @xmath6 to canonical principal parameters .
moreover , the function @xmath41 that defines @xmath6 via the ganchev s formula is given by @xmath42.@xmath39 the canonical principal parameters @xmath19 are determined uniquely up to the changes @xmath43
by investigating minimal bzier surfaces cosn and monterde @xcite formulate that any bi - cubic minimal surface defined by @xmath44 is , up to affine transformation in the space , actually an affine reparametrization of the classical enneper surface @xmath45 ( this chart of the enneper surface is obtained from the weierstrass formula with @xmath46 , @xmath47 . )
we note that their proof actually refers to bi - cubic * isothermal * minimal charts . indeed , as we mentioned in section [ preliminaries ] , the change of parameters to isothermal ones is in general nonlinear .
below we give a simple example of a bi - cubic minimal chart that can not be transformed by an affine transformation into an isothermal one .
consider the bi - cubic chart @xmath48 it can be shown by direct computation that this chart defines a minimal surface .
of course we can simply remark that this is a reparametrization of @xmath49 with @xmath1 replaced by @xmath50 , so the mean curvature vanishes identically .
let us make an affine transformation of the parameters @xmath19 : @xmath51 with nonzero jacobian , i.e. @xmath52 we shall try to determine the coefficients @xmath53 , so that the chart @xmath54 be isothermal .
actually we shall see what follows only from @xmath55 .
a direct computation shows that @xmath56 , where @xmath57 since @xmath58 is positive , the vanishing of @xmath59 implies @xmath60 .
hence the coefficients in @xmath61 must be zero .
in particular the coefficients of @xmath62 , @xmath63 , @xmath64 are @xmath65 these equations imply immediately @xmath66 let e.g. @xmath67 . from it follows @xmath68 , which contradicts .
so it is impossible to make an affine transformation of the parameters in to obtain an isothermal chart
. * remark .
* in view of the above notes the problem of existing bi - cubic minimal surfaces different from the enneper one is still open .
more generally it will be interesting to obtain a method for finding polynomial minimal non - isothermal charts .
in this section we examine minimal surfaces represented by isothermal polynomial charts of degree 4 in both @xmath69 .
we may expect that there exists more than one such surface , but it is interesting to know `` how many '' are there .
so consider the chart @xmath70 where @xmath71 are vectors in @xmath72 .
using @xmath21 and looking on its coefficient of @xmath73 we obtain @xmath74 .
analogously we derive consecutively @xmath75 , @xmath76 , @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 .
it is known that any minimal isothermal chart is harmonic , see e.g. @xcite . in our case
this implies @xmath84 substituting these in @xmath85 and looking on the coefficients of @xmath86 and @xmath87 we obtain also @xmath88 , @xmath89 .
if @xmath90 the chart is not of degree 4 .
so we assume @xmath91 . up to position in space and symmetry we may take @xmath92 now the coefficients of @xmath93 and @xmath94 in @xmath85 give @xmath95 and @xmath96 . using this
we can calculate the derivatives @xmath97 and @xmath9 .
let the functions @xmath22 and @xmath23 give the weierstrass representation of the surface .
denote by @xmath98 the derivative of @xmath99 .
then @xmath100 in our case a direct computation shows @xmath101 on the other hand the weierstrass formula implies easily @xmath102 hence we derive @xmath103 g(z)&=&\frac { c_{01 } + i c_{10 } + ( c_{11 } + 2 i c_{20 } + 3 ( - c_{03 } + i c_{30 } ) z ) z } { a_{01 } - i b_{01 } + i a_{10 } + b_{10 } + ( a_{11 } - i b_{11 } + 2 i a_{20 } + 2 b_{20 } ) z}\cdot\end{aligned}\ ] ] consequently we have obtained that for some complex constants @xmath104 and @xmath105 @xmath106 where @xmath107 is a polynomial of degree at most 2 .
suppose @xmath108 , i.e. @xmath22 is a constant .
then the derivative @xmath109 is of degree 2 or 4 , so the chart @xmath0 is of degree 3 or 5 , which is not our case .
so @xmath110 .
since @xmath111 is a polynomial then @xmath112 divides @xmath107 .
hence @xmath113 , where @xmath114 .
we have proved the following [ th1 ] any bi - quartic parametric polynomial minimal surface in isothermal parameters is generated by the weierstrass formula with the functions @xmath115 further , we are interested which of the functions in theorem [ th1 ] generate different surfaces .
denote by @xmath116 the chart defined as the real part of the weierstrass minimal curve with functions @xmath117 , @xmath47 and the corresponding surface by @xmath118 .
denote also by @xmath0 the chart with generating functions @xmath119 , @xmath120 for arbitrary nonzero complex numbers @xmath104 , @xmath121 , and the corresponding surface by @xmath6 . using and
we can see that the nonzero coefficients of the first fundamental form and the normal curvature of @xmath116 are respectively @xmath122 obviously @xmath116 is not in canonical principal parameters .
we want to change the parameters @xmath19 to canonical principal ones .
equation has a solution @xmath123 so according to theorem b we change the variable @xmath124 by @xmath125 .
now the functions @xmath126 generate a chart @xmath127 in canonical principal parameters and @xmath128 analogously @xmath0 is not in canonical principal parameters . using again theorem b and changing the complex variable @xmath124 by @xmath129 we obtain a corresponding chart @xmath130 in canonical principal parameters . according to its normal curvature is @xmath131 the last formula implies that this is also the normal curvature ( in canonical principal parameters ) of the surface @xmath132 generating via the weierstrass formula by the functions @xmath133 according to theorem a the surfaces @xmath6 and @xmath132 coincide up to position in the space . on the other hand , the weierstrass formula implies that the surfaces @xmath118 and @xmath132 are homothetic .
so for any nonzero complex numbers @xmath104 , @xmath105 the surface @xmath6 is , up to position in the space , homothetic to @xmath118 .
surfaces of type @xmath6 for different values of @xmath104 and @xmath121 are shown in figure [ generating ] . for different values of @xmath104 and @xmath121,scaledwidth=80.0% ] generating functions f(z)=10z , g(z)=z for different values of @xmath104 and @xmath121,scaledwidth=80.0% ] generating functions f(z)=z , g(z)=10z now we are interested whether the functions @xmath117 , @xmath134 , @xmath135 , define a surface which is really different from @xmath118 .
the chart @xmath0 generated by these functions is not in canonical principal parameters so according to theorem b we change the complex variable @xmath124 by @xmath125 .
then the functions @xmath136 define a chart @xmath130 in canonical principal parameters .
its normal curvature is @xmath137 where @xmath138 is @xmath139 comparing these functions for different values of @xmath140 we can say that the resulting surfaces are different . in figure [ comparison ]
are shown parts of the surface @xmath118 , obtained for @xmath141 ( left ) , the surface obtained for @xmath142 ( center ) , and the surface obtained for @xmath143 ( right ) . and
@xmath134,scaledwidth=90.0% ] @xmath144 and @xmath134,scaledwidth=90.0% ] @xmath145 and @xmath134,scaledwidth=90.0% ] @xmath146
we consider bi - quartic tensor product bzier surface defined by @xmath147 where @xmath148 are the control points of @xmath0 , and @xmath149 are the bernstein polynomials of degree @xmath150 defined for @xmath151 by @xmath152 recall that if @xmath0 is in isothermal parameters then @xmath0 is a minimal surface if and only if @xmath0 is a harmonic surface , i.e. @xmath153 . for a harmonic bzier surface monterde @xcite has proved that if we know the control points on two opposite boundaries except one corner point , e.g. nine points @xmath154 and @xmath155 , then the remaining sixteen control points are fully determined .
the proof , where @xmath156 is even .
here we consider the case @xmath157 .
] is based on the harmonic condition @xmath153 which leads to a linear system that has a unique solution .
@xmath158 { \bf b}_{30}&\circ & \circ & \circ & { \bf b}_{34}\\[1ex ] { \bf b}_{20}&\circ & \circ & \circ & \circ\\[1ex ] { \bf b}_{10}&\circ & \circ&\circ & { \bf b}_{14}\\[1ex ] { \bf b}_{00}&\circ & \circ & \circ & { \bf b}_{04 } \end{array } $ ] here
we assume that nine control points @xmath159 , @xmath160 as shown on figure [ f2 ] are given .
note that they differ from those used in @xcite . in lemma
[ lemma1 ] below we give expressions for the remaining control points through the given points .
then in proposition [ pr1 ] we prove that in the case where the given points are symmetric with respect to any of the coordinate planes , then the corresponding harmonic bzier surface is symmetric with respect to the same coordinate plane .
[ lemma1 ] let nine control points @xmath161 ; @xmath162 be given .
a bi - quartic bzier surface is harmonic if and only if the remaining sixteen control points satisfy @xmath163 * proof . *
it follows straightforward using the corresponding linear system from @xcite .
@xmath39 [ pr1 ] let the given points @xmath161 ; @xmath162 be symmetric with respect to some of the coordinate planes @xmath3
, @xmath4 , or @xmath5 .
then the corresponding harmonic bzier surface defined by lemma [ lemma1 ] is symmetric with respect to the same plane .
* let @xmath164 and assume that the given control points are symmetric with respect to the plane @xmath3 , i.e. @xmath165 and @xmath166 are symmetric points to @xmath167 and @xmath168 , respectively , @xmath169 , and @xmath170 lies on @xmath3 .
then we have @xmath171 for @xmath169 , and @xmath172 . to show that the harmonic bzier surface defined by lemma [ lemma1 ] is symmetric with respect to @xmath3 it suffices to establish that its control points are symmetric with respect to @xmath3 .
we need to establish that holds for @xmath173 and @xmath174 for @xmath175 .
next we verify that @xmath176 , @xmath177 , @xmath178 , and @xmath179 .
the analogous relations for the remaining control points follow in a similar way . from
and we have @xmath180 analogous relation holds for @xmath181 and @xmath182 . for
the third coordinates @xmath183 and @xmath184 we obtain @xmath185 it remains to show that @xmath179 .
we have @xmath186 the case where the nine given points are symmetric with respect to the other coordinate planes is treated analogously .
@xmath39 a bi - quartic harmonic bzier surface which is symmetric with respect to @xmath4 is shown from two different viewpoints in figure [ bezier1 ] .
its control points are presented in table [ tab1 ] .
we note that they are obtained from the minimal bi - quartic bzier surface with generating functions @xmath117 , @xmath187 .
hence , the surface in figure [ bezier1 ] is harmonic minimal bzier surface .
ccccc + @xmath188 & @xmath189&@xmath190&@xmath191 & @xmath192 + @xmath193&@xmath194&@xmath195&@xmath196 & @xmath197 + @xmath198&@xmath199&@xmath200&@xmath199 & @xmath201 + @xmath202&@xmath203&@xmath195&@xmath204&@xmath205 + @xmath206&@xmath207&@xmath208&@xmath209&@xmath210 + .
the surface is shown from two different viewpoints . ] .
the surface is shown from two different viewpoints . ]
in this paper we characterize all bi - quartic parametric polynomial minimal surfaces by their generating functions using the weierstrass formula .
we also consider the bi - quartic harmonic bzier surfaces and establish their symmetry with respect to any of the coordinate planes .
we present numerical experiments and give examples . a possible direction for future work is to extend our results for minimal surfaces of higher degrees .
ognian kassabov + department of mathematics and informatics + university of transport `` todor kableshkov '' + 158 g. milev str .
+ 1574 sofia , bulgaria + email : ` [email protected] ` + krassimira vlachkova + faculty of mathematics and informatics + sofia university + 5 james bourchier blvd .
+ 1164 sofia , bulgaria + email : ` [email protected] ` + | minimal surfaces with isothermal parameters admitting bzier representation were studied by cosn and monterde .
they showed that , up to an affine transformation , the enneper surface is the only bi - cubic isothermal minimal surface .
here we study bi - quartic isothermal minimal surfaces and establish the general form of their generating functions in the weierstrass representation formula .
we apply an approach proposed by ganchev to compute the normal curvature and show that , in contrast to the bi - cubic case , there is a variety of bi - quartic isothermal minimal surfaces .
based on the bzier representation we establish some geometric properties of the bi - quartic harmonic surfaces .
numerical experiments are visualized and presented to illustrate and support our results . _
key words _ : minimal surface , isothermal parameters , bzier surface |
the control of chaos in nonlinear systems has been an active field of research due to its potential applications in many practical situations where chaotic behaviour is not desirable . since the time of grebogi - ott - yorke @xcite
several methods were proposed to control chaotic dynamics to desired periodic states or to stabilize unstable fixed points @xcite of the system .
recently these techniques have been extended to control spatio - temporal chaos in spatially extended systems @xcite .
they are useful in many applications to control dynamics in plasma devices and chemical reactions @xcite and to reduce intensity fluctuations in laser systems @xcite .
we note that control is important for emergence of regulated and sustainable phenomena in biological systems , for example to ensure stability of signal - off state in cell signaling networks which is desirable to prevent autoactivation @xcite .
also in general , chaotic oscillations can degrade the performance of engineered systems and hence effective and simple control strategies have immense relevance in such cases .
the dynamics of spatially extended systems can be modeled in a very simple but effective manner by using coupled map lattices ( cml ) introduced by kaneko @xcite .
this approach forms an efficient method to coarse grain the local dynamics in such systems and has been used to understand complex natural phenomena .
the dynamical states in such systems are extremely rich and varied , including spatiotemporal chaos , regular and irregular patterns , travelling waves , spiral waves etc.@xcite . in the case of coupled map lattices , methods for control and synchronization reported earlier
are mostly based on nonlinear feedback control @xcite , constant and feedback pinning , etc . @xcite .
recently coupled chaotic maps have been shown to stabilize to a homogeneous steady state due to time delay in coupling @xcite and in the presence of random delays @xcite .
similarly , using multiple delay feedback , unstable steady states are shown to stabilize in chaotic electronic circuits @xcite . under threshold activated coupling at selected pinning sites , chaotic neuronal maps are reported to stabilize to regular periodic patterns @xcite . in one way coupled map lattice decentralized delayed feedback
can introduce control of chaos @xcite .
we note that most of the control schemes like feedback and pinning , the control units are often derived from the dynamics of the system and as such must be designed specific to each system . also delay feedback makes the system higher dimensional from the analysis point of view and requires careful choice of delay and its implementation for achieving control .
however in many applications for the practical implementation of the control scheme , it is desirable to have a general scheme requiring minimum information about the system to be controlled . in this paper
, we introduce a coupling scheme , which can be applied externally to the system and does not require a priori information about the system
. as such it is very general and effective and can be implemented easily for control of spatio - temporal dynamics and patterns on coupled discrete systems .
we note that in the context of continuous systems , one of the methods recently reported to induce amplitude death or steady state in coupled systems is coupling to an external damped system referred to as environmental coupling or indirect coupling @xcite .
this method has been successfully implemented using electronic circuits @xcite and applied for controlling dynamics of single systems @xcite and systems with bistability @xcite as linear augmentation .
these methods in general involve a single external system to control the dynamics of coupled continuous systems .
the present paper extends this particular method in two ways : the method proposed is for controlling dynamics in discrete systems and the external control system is a spatially extended system .
we find that the spatial extension has its own advantages and relevance as a method for suppressing dynamics .
the control system in our scheme is designed as an external lattice of damped discrete systems such that without feedback from the system , this control lattice stabilizes to a global fixed point . when system and control lattices are put in a feedback loop ,
the mutual dynamics works to control their dynamics to a homogeneous steady state or periodic state .
we illustrate this using logistic map and henon map as site dynamics .
the analysis is developed starting with a single unit of the interacting system and the control , which is then extended to connected rings of systems , interacting 2-dimensional lattices and random networks .
this ` bottom up approach ' is chosen , not only because it gives clarity in describing the mechanism but also because the cases of even single system or rings are not studied or reported so far for discrete systems .
we analyze the stability of the coupled system and control lattices using the theory of circulant matrices and routh - hurwitz s criterion for discrete systems .
thus we obtain regions in relevant parameter planes that correspond to effective control .
we also report results from detailed numerical simulations that are found to agree with that from the stability analysis . in particular cases ,
we obtain control even when the units in the system lattice are uncoupled .
so also , uncoupled units in the control lattice can control the coupled system lattice .
moreover by tuning the parameters of the control lattice , we can achieve control to regular periodic patterns on the system lattice .
recently , chimera states in a 1-d lattice with nonlocal couplings have been realized using liquid crystal light modulator @xcite .
we show how our scheme can control the chimera states in this system .
the extended and external control system introduced here can model an external medium effectively in controlling the dynamics of real world systems . as an example
, we show how the dynamical patterns and undesirable excitations produced by coupled neurons can be controlled due to interaction with the extra cellular medium . in the end
we extend the scheme to control the dynamics of discrete time systems on a random network . in this case
, the stability of the controlled state is analyzed using master stability function method and supported by direct numerical simulations .
in this section we introduce our scheme for controlling a 1-d cml of size @xmath0 by coupling with an equivalent lattice of damped systems .
the dynamics at the @xmath1 node of the system lattice constructed using the discrete dynamical system , @xmath2 , with time index @xmath3 , is given by : @xmath4 here @xmath5 @xmath6 @xmath7 is the state vector of the discrete system , @xmath8 is the nonlinear differentiable function . also , @xmath9 represents the strength of diffusive coupling and @xmath10 is a @xmath11 matrix , entries of which decide the variables of the system to be used in the coupling .
the dynamics of a damped discrete map @xmath12 in the external lattice is represented as : @xmath13 where k is a real number with @xmath14 .
thus for positive values of @xmath15 , this map is analogous to over - damped oscillator while for negative values of @xmath15 it is analogous to damped oscillator .
the dynamics at the @xmath1 node of the 1-d cml of size @xmath0 of these maps with diffusive coupling can be represented by the following equation : @xmath16 we have , @xmath17 and @xmath18 represents the strength of diffusive coupling among the nodes of the external lattice .
we consider periodic boundary conditions for both the lattices .
when the system lattice is coupled to the lattice of external systems , node to node , with feedback type of coupling , then the resulting system can be represented as : @xmath19 here @xmath20 is the @xmath21 matrix which determines the components of the state vector @xmath5 which take part in the coupling with external system . in the next section
we start with the analysis of control for one unit of this coupled system
. then we study the case of 1-d cml and 2-d cml and extend the scheme to random networks .
in all these cases , for direct numerical simulations , the typical discrete systems considered are the logistic map and the henon map .
the dynamics of a single unit of the model introduced in eq.([coupled1dcml ] ) is given by a discrete system coupled to an external system in a feedback loop as : @xmath22 to proceed further we consider logistic map as system dynamics to get : @xmath23 the parameter @xmath24 whose value determines the intrinsic nature of dynamics of logistic map is chosen as @xmath25 , so that the map is in the chaotic regime .
time series @xmath26 obtained from eq.([singleunitlogistic ] ) numerically is plotted in fig .
[ logisticts ] for two different values of coupling strength with @xmath27 .
the control is switched on at @xmath28 and we find that depending on the coupling strength , the chaotic dynamics can be controlled to periodic or steady state . of the system in eq.([2dlogistic ] ) for 2 different values of feedback coupling strength with @xmath29 and @xmath27 .
initially the logistic map is in the chaotic regime and the coupling with external system is switched on at time @xmath30.(a ) for @xmath31 , chaotic dynamics is controlled to periodic dynamics .
( b ) for @xmath32 , the controlled state is a fixed point .
] we find that there are @xmath33 fixed points for the system in eq.([singleunitlogistic ] ) given by the following equations : @xmath34 and @xmath35 to analyze the stability of these fixed points , we consider jacobian of the system in ( [ 2dlogistic ] ) evaluated at the fixed point . @xmath36 the characteristic equation for this matrix is given by : @xmath37 where @xmath38 the corresponding fixed point will be stable if the absolute values of all the eigenvalues given by eq.([characteristiclogistic ] ) are less than @xmath39 .
using routh - hurwitz s conditions for discrete systems , this would happen when the following conditions are satisfied @xcite : @xmath40 using these conditions , we find that the fixed point @xmath41 is unstable . the other fixed point given by eq.([logisticfixednontrivial ] ) is stable for a range of parameter values and these ranges can be obtained using the analysis described above .
the numerical and analytical results thus obtained are shown in fig .
[ logisticplanes ] . from fig .
[ logisticplanes](a ) it is clear that the steady state region ( red ) is symmetric about the line @xmath42 and the line @xmath27 .
[ logisticplanes](b ) shows the region of steady state behavior in @xmath43-@xmath15 plane where @xmath44 . in both these plots ,
black boundary is obtained analytically and green region , the region in which system is oscillatory , is obtained numerically .
it is clear that curves in black on the boundaries of the region that correspond to steady state , obtained analytically using eq.([stabilityconditions2d ] ) , agree well with the results of numerical analysis .
( color online ) parameter planes for a chaotic logistic map coupled to an external map : ( a ) @xmath45-@xmath46 plane for @xmath47 , ( b ) @xmath43-@xmath15 plane where @xmath44 . here
, the region where the fixed point of coupled system is stable is shown in red ( dark gray ) , in the green ( light gray ) region system is oscillatory while in the white region system is unstable .
the black boundary surrounding the red region , in both the plots , is obtained analytically . ]
we observe that the transition to steady state behaviour for this system , for a given value of @xmath15 as @xmath43 is varied , happens through reverse doubling bifurcations .
this would mean that by tuning @xmath43 , it is possible to control the system to any periodic cycle or fixed point .
as a second example we consider the henon map as the site dynamics .
it is a 2-d map given by : @xmath48 then the dynamics of the coupled system is given by : @xmath49 the intrinsic dynamics of henon map is chaotic with parameters @xmath50 and @xmath51 .
it is clear from time series given in fig .
[ henonts ] that the coupled system in eq.([3dhenon ] ) , can be controlled to periodic state or fixed point state by varying the parameters @xmath45 , @xmath46 and @xmath15 .
time series , @xmath26 of the system in eq.([3dhenon ] ) for @xmath33 different values of coupling strength with @xmath27 and @xmath29 .
initially henon map is in the chaotic regime and the coupling with external system is switched on at time @xmath52 . ( a )
@xmath53 , chaotic dynamics is controlled to periodic dynamics of period two , ( b ) @xmath54 , the system is controlled to a fixed point . ] the @xmath55-d dynamical system in ( [ 3dhenon ] ) has a pair of fixed points given by : @xmath56\nonumber\\ y^{\ast}&=&\frac{b}{2a}\left[-c\pm \sqrt{c^{2}+4a}\right]\\ z^{\ast}&=&\frac{\varepsilon}{2a(1-k)}\left[-c\pm \sqrt{c^{2}+4a}\right]\nonumber\end{aligned}\ ] ] where , @xmath57 the stability of these fixed points is decided by the jacobian matrix @xmath58 evaluated at that fixed point : @xmath59 the characteristic polynomial equation for the eigenvalue @xmath60 , in this case , is given by : @xmath61 where @xmath62 similar to the case of logistic map , the absolute values of all eigenvalues given by eq.([characteristichenon ] ) are less than @xmath39 if the following conditions are satisfied @xcite : @xmath63 here also , one of the solutions in eq.([3dhenonfixedpoints ] ) ( with negative sign ) is unstable while the other is stable for regions in parameter planes @xmath45-@xmath46 and @xmath43-@xmath15 as shown in fig . [ henonplanes ] .
the color codes and details are as given in fig .
[ logisticplanes ] . for any particular value of parameter @xmath15 ,
the nature of transition in this case is through a sequence of reverse period doubling bifurcations similar to the case of logistic map .
( color online ) parameter planes for a coupled system of henon map and external map .
( a)@xmath45-@xmath46 plane for @xmath29 , ( b)@xmath43-@xmath15 plane .
color code is same as that for fig .
[ logisticplanes ] ] in this section , we consider control of dynamics on a ring of discrete systems . for this , as mentioned in section ii , we construct a ring of external maps which are coupled to each other diffusively and then couple it to the ring of discrete systems in one - to - one fashion with feedback type of coupling . with logistic map as on - site dynamics , the dynamics at the @xmath1 site of the coupled system is given by : @xmath64 where @xmath65 numerically , we find that the system in eq.([coupledlogisticrings ] ) can be controlled to a periodic state and a state of fixed point or amplitude death by adjusting the coupling strengths @xmath45 and @xmath46 . fig .
[ space - timelogisticring ] shows two such cases , where in ( a ) lattice is controlled to a temporal @xmath33-cycle and in ( b ) temporal fixed point state .
( color online ) space - time plots for a ring of logistic maps coupled to an external ring .
the coupling is switched on at time @xmath66 .
parameter values of the system are : @xmath25 , @xmath67 , @xmath68 , @xmath47 .
( a ) for @xmath69 ; @xmath70 , the dynamics is controlled to a @xmath33-cycle state and ( b ) for @xmath71 ; @xmath72 , the dynamics is controlled to a fixed point state .
the color code is as per the value of @xmath26 ] to analyze the stability of the controlled and synchronized fixed point , we consider the system in eq.([coupledlogisticrings ] ) as a 1-d lattice of single units considered in section iii coupled diffusively through x and z. because of the synchronized nature of the fixed point and periodic boundary conditions , jacobian of this system is block circulant matrix as given below .
@xmath73 for the case of logistic maps , @xmath74 and @xmath75 are @xmath76 matrices and @xmath77 denotes @xmath76 zero matrix . in explicit form , these matrices are given by : @xmath78\end{aligned}\ ] ] and @xmath79\end{aligned}\ ] ] following the analysis given in @xcite for eigenvalues of the general block circulant matrix as given below : @xmath80 we construct @xmath0 blocks each of order @xmath76 as follows : @xmath81 where @xmath82 is a complex @xmath0th root of unity : @xmath83 with @xmath84 + for our problem , @xmath85 and @xmath86 .
thus we get , @xmath87 where @xmath88 thus , in this case , @xmath89 is given by : @xmath90 where , @xmath91 now we make use of the fact that eigenvalues of jacobian when evaluated at the synchronized fixed point are the same as eigenvalues of these @xmath89 matrices since the jacobian is block circulant matrix @xcite .
this means that , to check the stability of the fixed point of the two coupled rings , instead of directly calculating eigenvalues of jacobian matrix of size @xmath92 , we can calculate eigenvalues of all @xmath89 matrices , each of which is @xmath76 matrix , which is a much simpler task .
the characteristic equation for @xmath89 is given by : @xmath93 where @xmath94 in this case , the corresponding fixed point with coordinates given by eq.([logisticfixedtrivial ] ) or eq.([logisticfixednontrivial ] ) is stable when the conditions given in eq.([stabilityconditions2d ] ) are satisfied for each @xmath89 .
these conditions then give us the regions in parameter space where the fixed point state of the whole system is stable i.e. the regions where the control of the original dynamics to the steady state is possible .
similar to the case of single unit , the fixed point with coordinates given by eq.([logisticfixedtrivial ] ) turns out to be unstable for all values of control parameters .
the other fixed point is stable for a range of parameter values .
these amplitude death regions obtained numerically in @xmath43-@xmath15 and @xmath9-@xmath18 planes for this system of coupled rings are shown in fig .
[ logisticringplanes ] . in each of these plots ,
the amplitude death region is shown with red color while the black boundary around that region is obtained analytically . from fig .
[ logisticringplanes](b ) , it is clear that the amplitude death can happen even when the logistic maps are not coupled to each other directly ( corresponding to @xmath95 ) or when external maps are not coupled to each other ( corresponding to @xmath96 ) .
( color online ) parameter planes for a ring of logistic maps coupled to an external ring : ( a ) @xmath43-@xmath15 plane when @xmath97 and @xmath98 , ( b ) @xmath9-@xmath18 plane when @xmath99 and @xmath47 .
color code is same as that for fig .
[ logisticplanes ] ] as our next example , we illustrate the control scheme for a ring of henon maps .
the coupled system in this case can be represented as follows : @xmath100 where @xmath101 here also the system can be controlled to a synchronized fixed point and because of the periodic boundary conditions and because of the synchronized nature of fixed point , jacobian becomes block circulant as given in eq.([jacobian ] ) . in this case ,
@xmath74 and @xmath75 are @xmath102 matrices and @xmath77 represents @xmath102 zero matrix . in explicit form , @xmath74 and @xmath75 are given by : @xmath103 and @xmath104 let @xmath105 then , @xmath106 the characteristic equation for @xmath89 is given by : @xmath107 where @xmath108 using these parameters @xmath109 , we define @xmath110 and @xmath111 as in the case of single unit in eq.([stabilityconditions3d ] ) .
this allows us to check stability for each of the @xmath89 matrices which in turn gives us regions in parameter space where the control of the dynamics is possible for a ring of henon maps . in fig .
[ epskhenonring ] we show these regions in @xmath43-@xmath15 parameter plane , for @xmath96 and for @xmath95 .
( color online ) @xmath43-@xmath15 parameter planes for a ring of henon maps coupled to an external ring : ( a ) when @xmath97 , @xmath96 , ( b ) when @xmath95 , @xmath98 .
color code is same as that for fig .
[ logisticplanes ] ] as an application of our scheme , we present the control of chimera states produced on a 1-d lattice with non - local couplings .
the chimera states have been observed in many spatially extended systems and coupled oscillators with non - local couplings @xcite .
these states , which correspond to the coexistence of spatial regions with synchronized and incoherent behavior , may not be desirable in many systems like power grids .
here we specifically consider the chimera states experimentally realized using liquid - crystal spatial light modulator @xcite .
this system corresponds to a 1-d lattice with each node diffusively coupled to @xmath112 of its nearest neighbours . in this case
the phase of each node @xmath113 is the dynamical variable whereas a physically important quantity is the output intensity i , related to phase as : @xmath114 the coupled map lattice is then represented by the following set of equations @xcite : @xmath115 \right\}\end{aligned}\ ] ] for the coupling radius @xmath116 and @xmath117 as shown in @xcite , the spatial profile of intensity has two distinct synchronized domains which are separated by small incoherent domains confirming the existence of chimera states
. the coupling with external lattice in this case can be as : @xmath118 \right\}+\varepsilon_{1}z_{i}(n)\nonumber\\ z_{i}(n+1 ) & = & kz_{i}(n ) + d_{e}(z_{i-1}(n)+z_{i+1}(n)-2z_{i}(n ) ) + \varepsilon_{2}\phi_{i}(n ) \end{aligned}\ ] ] we find that with this control scheme , spatial as well as temporal control of chimera state can be achieved .
we plot numerically the chimera states obtained numerically from eq.([slmcml ] ) and the controlled state from eq.([slm_control ] ) in fig .
[ slm](a ) . fig .
[ slm](b ) shows time series of a typical node indicating a control to steady state .
( a ) chimera state developed in a 1d coupled map lattice of size @xmath119 represented by eq.([slmcml ] ) shown with red color . after the control is applied , chimera state is replaced by a spatially synchronized state as shown in black .
( b ) the time series of a typical node of the cml .
when chimera state is present , individual node is in 2-cycle state . at time
@xmath52 control is switched on with @xmath47 , @xmath120 which decreases the intensity fluctuations though 2-cycle is still present .
further increase in coupling with @xmath121 at time @xmath122 gets rid of all fluctuations and the intensity goes to a constant value spatially and temporally . ]
now we consider controlling the dynamics on 2-d lattice of discrete systems by coupling to an external lattice as follows : @xmath123 here , @xmath20 is the @xmath21 matrix which determines the components of state vector @xmath5 which take part in the coupling with an external system . in fig .
[ logisticlatticets ] , we show the numerically obtained time series from a typical node of a @xmath124 lattice of logistic maps coupled to an external lattice for @xmath55 different coupling strengths @xmath43 . in all three cases , coupling with an external lattice
is switched on at time @xmath125 .
the time series @xmath26 of a typical node from a @xmath124 lattice of logistic maps coupled to an external lattice when @xmath27 .
parameter values are : @xmath67 , @xmath68 and @xmath25 .
in all the cases coupling with external lattice is switched on at time @xmath28 .
( a ) for @xmath126 , the chaotic dynamics is controlled to 4-cycle state .
( b ) for @xmath70 , it is a 2-cycle state .
( c ) for @xmath72 , the dynamics is quenched to fixed point state or amplitude death state . ] in fig .
[ logisticlatticest ] , we show space - time plots of 2-d lattice of logistic maps with and without coupling with an external lattice . here the time series from nodes along one of the main diagonals of the lattice are plotted on the y - axis and time on the x - axis .
the coupling strength with the external lattice is adjusted so that lattice of logistic maps is controlled to a 2-cycle state temporally .
( color online ) ( a ) space - time plot of patterns formed on 2-d lattice of logistic maps when the individual maps are in chaotic regime , @xmath25 and @xmath67 ( b ) when this lattice is coupled to the external lattice with @xmath69 , @xmath70 , @xmath68 , the dynamics is controlled to a 2-cycle state temporally .
also , spatially pattern on the lattice becomes regular .
color coding in the plot is according the values of @xmath26.,height=226 ] the stability of the synchronized steady state of coupled 2-d lattices can be analyzed in exactly similar fashion to the coupled rings discussed in detail in previous section by re - indexing lattice site @xmath127 with a running index @xmath128 defined as @xmath129 .
then jacobian for the whole system becomes @xmath130 block circulant matrix with each block s order being @xmath131 .
the intersection of regions satisfying the stability conditions for all the @xmath132 matrices will then give the region of amplitude death in the corresponding parameter planes .
the results thus obtained are qualitatively similar to the case of the ring and therefore not included here . in this section
we show how the present scheme can provide a phenomenological model for the role of external medium in controlling the dynamics of coupled neurons .
it is established that neurons can cause fluctuations in the extra cellular medium which when fed back to the neurons can affect their activity via ephaptic coupling @xcite . in this context , we take the 2-d control lattice as the discrete approximation to the medium and the dynamics of individual neuron , to be the rulkov map model@xcite given by the following equations : @xmath133 here @xmath134 , represents the membrane potential of the neuron and @xmath135 is related to gating variables .
@xmath136 , @xmath137 and @xmath138 are intrinsic parameters of the map .
we take @xmath139 and for @xmath140 , there exists a stable fixed point for the map given by @xmath141 and @xmath142 . in the range @xmath143 to @xmath144
, the map shows periodic behaviour .
we model the system of coupled neurons as a 2-d lattice which is in a feedback loop with the external lattice as given by : @xmath145 for a realistic modeling of the system , we take the values of the parameter @xmath136 distributed uniformly and randomly between @xmath146 and @xmath147 among all the maps and the extra cellular medium as the external lattice of same size with a spread in @xmath15 values distributed randomly in @xmath148 $ ] . in the absence of coupling with medium , the neurons can form dynamical patterns of the type shown in fig .
[ pattern](a ) obtained by iterating the lattice for @xmath149 time steps with @xmath150 and initial conditions for @xmath134 and @xmath135 randomly and uniformly distributed in @xmath151 $ ] .
when coupling with the external lattice is switched on , the patterns get suppressed giving steady and synchronized states . in fig .
[ pattern](b ) we show the space time plot for the lattice with and @xmath152 , @xmath153 and @xmath154 and the coupling is switched on at time @xmath155 .
the horizontal axis represents the systems on the main diagonal of the lattice and vertical axis is time .
as is clear from the plot , the external lattice suppresses the pattern on the lattice .
we note that dynamical spatio - temporal patterns can arise in the neuronal lattice because of defect neurons that are different from others .
we illustrate this by considering a lattice of neurons in which all neurons have @xmath136 in the range @xmath156 to @xmath157 except one neuron which is in the excited state with @xmath158 .
the patterns produced are shown in fig .
[ defect](a ) .
such patterns of propagating excitation waves can be pathological unless suppressed .
coupling with the external medium can effectively suppress them .
[ defect](b),(c),(d ) display the snapshots of the lattice after coupling with the external lattice .
now we generalize the control scheme given above to control the dynamics on random networks of discrete systems . in this case
, we construct external network with the same topology as the system network and then we couple these networks in one to one fashion with feedback type of coupling .
this coupled system of two networks , when the coupling between the nodes of the same network is of diffusive type , can be represented by : @xmath159 in this equation , intrinsic @xmath160-dimensional dynamics on the nodes of the network is given by @xmath161 and the same function @xmath162 is used to couple nodes in the network diffusively . also , @xmath163 is the @xmath164 entry of the adjacency matrix @xmath165 of the network , @xmath10 is the @xmath11 matrix which determines the components of state vector @xmath5 to be used in the coupling with the other nodes and @xmath166 is the @xmath21 matrix which determines the components of state vector @xmath5 to be used in the coupling with nodes of the external network .
we perform numerical simulations for a network of @xmath125 nodes for @xmath167 different random topologies which include three realizations of erds - rnyi network with average degree equal to @xmath168 and two realizations of barab ' asi - albert scale - free network .
the dynamics on the nodes is of logistic map in all these simulations with @xmath44 .
we find that for all topologies , as the coupling strength @xmath43 is increased both networks reach synchronized fixed point of the single unit given in eq.([singleunitlogistic ] ) .
this synchronized steady sate or fixed point can be considered as amplitude death state of the network and to analyze its stability , we invoke the master stability analysis for this fixed point @xcite .
since the external network has exactly the same topology as the original network and since nodes of two networks are coupled in one to one fashion , this system of two coupled networks can be considered as a single network of the single units considered in section iii .
then if @xmath169 represents @xmath170th component of perturbation to the synchronized fixed point of the network at node @xmath171 , we expand this perturbation as : @xmath172 where @xmath173 is @xmath1 component of @xmath24th eigenvector of laplacian matrix of the network . following the analysis given in @xcite
we then get : @xmath174{\bm c}^{r}(n)\ ] ] for logistic map as nodal dynamics and for diffusive coupling among the nodes , matrices @xmath175 and @xmath176 are given by : @xmath177 and @xmath178 this gives , @xmath179\ ] ] then for the synchronized fixed point state of the system of coupled networks to be stable , for every eigenvalue @xmath180 of the laplacian matrix of the network , absolute values of all the eigenvalues of the matrix @xmath181 should be less than @xmath39 .
we now consider @xmath180 in @xmath181 as a parameter @xmath60 and for a range of @xmath60 , find eigenvalue with largest absolute value and use this eigenvalue as master stability function ( msf ) .
the results are shown in fig .
[ msf ] where for @xmath55 different values of @xmath43 , the master stability curves are shown .
we see that all the curves are continuous over the whole range of @xmath60 .
the region of the curve for which @xmath182 is the stable region for that curve .
if all the eigenvalues of laplacian matrix for a given topology fall inside this region , then the state of synchronized fixed point is stable . since the smallest eigenvalue of laplacian matrix for unweighted and undirected network is always @xmath77 , we need to worry about the largest eigenvalue only .
hence if both @xmath183 and @xmath184 are inside this region then fixed point for the network is stable .
the red curve ( thick continuous ) in fig . [ msf ] is for @xmath185 which is the minimum coupling strength required for any topology for fixed point of the whole network to be stable . for any coupling strength less than this , absolute value of @xmath183 is greater than @xmath39 and hence fixed point of the network can not be stable . to illustrate this ,
we plot the master stability curve for @xmath186 shown in the figure as dotted blue curve .
the third curve ( thin continuous ) is for @xmath187 . in the figure we mark largest eigenvalues of laplacian matrices of @xmath167 different networks for which numerical simulations are performed .
( color online ) master stability functions for a network of logistic maps coupled to an external network for three values of coupling strengths @xmath43 .
the red curve ( thick curve ) is for critical value of @xmath185 which is the minimum coupling strength necessary for the fixed point state of the network to be stable when all other parameters are kept constant .
the blue curve ( dotted ) is for @xmath188 and it is clear from the figure that for all topologies , fixed point state is unstable at this coupling strength .
the black curve ( thin continuous ) is for @xmath187 . for a network of @xmath125 nodes , with different random topologies ( @xmath55 erds - rnyi topologies and @xmath33 scale - free topologies ) ,
the largest eigenvalue is marked on the graph . ] to support the master stability analysis , we present numerical results for one particular realization of erds - rnyi network .
we calculate an index @xmath189 to characterize the synchronized fixed point by taking the difference between global maximum and global minimum of the time series of @xmath134 at each node of the network after neglecting the transients and averaging this quantity over all the nodes . for parameter values corresponding to the controlled state of amplitude death
, this index has value @xmath77 @xcite . in fig .
[ adindexlogisticnetwork ] , we show variation of @xmath189 with @xmath43 and it is clear that @xmath189 goes to @xmath77 at @xmath185 indicating that amplitude death happens as predicted by master stability analysis .
the index @xmath189 obtained numerically as a function of coupling strength @xmath43 for a system of logistic network coupled to an external network .
the topology of both networks in this case is of erds - rnyi type . at approximately @xmath185 ,
control to steady state happens . ]
the suppression of unwanted or undesirable excitations or chaotic oscillations in systems is essential in a variety of fields and therefore mechanisms for achieving this suppression in connected systems are of great relevance . in the specific context of neurons , it is known that exaggerated oscillatory synchronization or propagation of some excitations can lead to pathological situations and hence methods for their suppression are very important .
methods to avoid oscillations by design of system controls such as power system stabilizers are important in engineering too . in this paper we present a novel scheme where suppression of dynamics can be induced by coupling the system with an external system .
our study is mostly on extended systems like rings and lattices in interaction with similar systems with different dynamics .
however the method is quite general and is applicable to other such situations also where control and stabilization to steady states are desirable . even in the case of single systems , as illustrated here for henon and logistic maps , this method can be thought of as a control mechanism for targeting the system to a steady state .
we develop the general stability analysis for the coupled systems using the theory of maps and circulant matrices and identify the regions where suppression is possible in the different cases considered .
the results from direct numerical simulations carried out using two standard discrete systems , logistic amp and henon map , agree well with that from the stability analysis .
our method will work even with a single external system as control ; however , extended or coupled external system enhances the region of effective suppression in the parameter planes .
our study indicates that by tuning the parameters of the external control system , we can regulate the dynamics of the original system to desired periodic behavior with consequent spatial order .
we also report how the present coupling scheme can be extended to suppress dynamics on random networks by connecting the system network with an external similar network in a feedback loop . in this context
the stability analysis using master stability function , gives the critical strength of feedback coupling required for suppression .
our analysis leads to the interesting result that control to steady state occurs even when the systems are not coupled among themselves but coupled individually to the connected external systems .
so also , individual external systems , which are not connected among themselves but each one connected to one node in the system , can suppress the dynamics of the connected nodes in the system .
this facilitates flexibility in the design of control through connected or isolated external systems depending on the requirement .
the scheme of external and extended control , introduced here , has relevance in understanding and modeling controlled dynamics in real systems due to interaction with external medium .
we illustrate this in the context of coupled neurons where the suppression of dynamical patterns is achieved by coupling with the extra cellular medium .
we also indicate how the same scheme can be applied to control chimera states in non - locally coupled systems . until recently ,
cml was considered as an idealized theoretical model for studies related to spatio - temporal dynamics .
however the recent works on a physically realizable cml system using liquid - crystal spatial light modulator @xcite and based on digital phaselocked loop @xcite have opened up many interesting possibilities in studying their control schemes also experimentally . in this context
we note that the scheme presented in the paper is suitable for a physical realization using optical instruments or electronic hardware .
one of the authors ( s.m.s . ) would like to thank university grants commission , new delhi , india for financial support .
00 ott e , grebogi c , yorke ja
. controlling chaos .
lett . 1990 ; 64:1196 - 1199 .
boccaletti s , grebogi c , lai y - c , mancini h , maza d. the control of chaos : theory and applications .
phys . rep .
2000 ; 329:103 - 197 .
schll e , schuster hg , editors .
handbook of chaos control .
wiley - vch ; 2007 .
klinger t , schrder c , block d , greiner f , piel a , bonhomme g , et al .
chaos control and taming of turbulence in plasma devices .
plasmas 2001 ; 8:1961 - 1968 .
crdoba a , lemos mc , jimnez - morales f. periodical forcing for the control of chaos in a chemical reaction .
2006 ; 124:014707 .
roy r , murphy tw jr , maier td , gills z , hunt er . dynamical control of a chaotic laser : experimental stabilization of a globally coupled system .
1992 ; 68:1259 .
wei m - d , lun j - c .
amplitude death in coupled chaotic solid - state lasers with cavity - configuration - dependent instabilities .
phys . lett . 2007 ; 91:061121 .
kondor d , vattay g. dynamics and structure in cell signaling networks : off - state stability and dynamically positive cycles .
plos one 2013 ; 8(3):e57653 .
kaneko k. spatiotemporal chaos in one- and two - dimensional coupled map lattices .
physica d 1989 ; 37:60 - 82 .
kaneko k. overview of coupled map lattices .
chaos 1992 ; 2:279 - 282 .
kaneko k , editor .
theory and applications of coupled map lattices .
wiley ; 1993 .
fang j , ali mk .
nonlinear feedback control of spatiotemporal chaos in coupled map lattices .
discrete dynamics in nature and society 1997 ; 1:283 - 305 .
parmananda p , hildebrand m , eiswirth m. controlling turbulence in coupled map lattice systems using feedback techniques .
phys . rev .
e 1997 ; 56:239 - 244 .
parekh n , parthasarathy s , sinha s. global and local control of spatiotemporal chaos in coupled map lattices .
1998 ; 81:1401 - 1404 .
gang h , zhilin q. controlling spatiotemporal chaos in coupled map lattice systems .
1994 ; 72:68 - 71 .
konishi k , kokame h. time - delay - induced amplitude death in chaotic map lattices and its avoiding control .
physics letters a 2007 : 366:585 - 590.s masoller c , marti ac .
random delays and the synchronization of chaotic maps .
2005 ; 94:134102 .
ahlborn a , parlitz u. stabilizing unstable steady states using multiple delay feedback control .
2004 ; 93:264101 .
shrimali md .
pinning control of threshold coupled chaotic neuronal maps .
chaos 2009 ; 19:033105 .
konishi k , kokame h. decentralized delayed - feedback control of a one - way coupled ring map lattice .
physica d 1999 ; 127:1 - 12 .
resmi v , ambika g , amritkar re .
general mechanism for amplitude death in coupled systems .
e 2011 ; 84:046212 .
resmi v , ambika g , amritkar re , rangarajan g. amplitude death in complex networks induced by environment .
e 2012 ; 85:046211 .
sharma a , sharma pr , shrimali md .
amplitude death in nonlinear oscillators with indirect coupling .
lett . a 2012 ; 376:1562 - 1566 .
banerjee t , biswas d. synchronization in hyperchaotic time - delayed electronic oscillators coupled indirectly via a common environment .
nonlinear dyn 2013 ; 73:2025 - 2048 .
sharma pr , sharma a , shrimali md , prasad a. targeting fixed - point solutions in nonlinear oscillators through linear augmentation .
e 2011 ; 83:067201 .
sharma pr , shrimali md , prasad a , feudel u. controlling bistability by linear augmentation .
lett . a 2013 ; 377:2329 - 2332 .
hagerstrom am , murphy te , roy r , hvel p , omelchenko i , schll e. experimental observation of chimeras in coupled - map lattices . nature physics 2012 ; 8:658 - 661 . sonis m. critical bifurcation surfaces of 3d discrete dynamics .
discrete dynamics in nature and society 2000 ; 4:333 - 343 .
eigenvectors of block circulant and alternating circulant matrices .
inf . math .
sci . 2005 ; 8:123 - 142 .
dudkowski d , maistrenko y , kapitaniak t. different types of chimera states : an interplay between spatial and dynamical chaos .
e 2014 ; 90:032920 .
panaggio mj , abrams dm .
chimera states : coexistence of coherence and incoherence in networks of coupled oscillators .
arxiv:1403.6204v2 [ nlin.cd ] 2014 .
yao n , huang z - g , lai y - c , zheng z - g .
robustness of chimera states in complex dynamical systems .
sci . rep .
2013 ; 3:3522 .
anastassiou ca , perin r , markram h , koch c. ephaptic coupling of cortical neurons .
nature neuroscience 2011 ; 14:217223 .
rulkov nf .
modeling of spiking - bursting neural behavior using two - dimensional map .
e 2002 ; 65:041922 .
pecora lm , carroll tl .
master stability functions for synchronized coupled systems .
phys . rev .
1998 ; 80:2109 - 2112 .
newman mej . networks .
oxford university press ; 2010 .
banerjee t , paul b , sarkar bc .
spatiotemporal dynamics of a digital phase - locked loop based coupled map lattice system .
chaos 2014 ; 24:013116 . | we present a new coupling scheme to control spatio - temporal patterns and chimeras on 1-d and 2-d lattices and random networks of discrete dynamical systems .
the scheme involves coupling with an external lattice or network of damped systems .
when the system network and external network are set in a feedback loop , the system network can be controlled to a homogeneous steady state or synchronized periodic state with suppression of the chaotic dynamics of the individual units .
the control scheme has the advantage that its design does not require any prior information about the system dynamics or its parameters and works effectively for a range of parameters of the control network .
we analyze the stability of the controlled steady state or amplitude death state of lattices using the theory of circulant matrices and routh - hurwitz s criterion for discrete systems and this helps to isolate regions of effective control in the relevant parameter planes .
the conditions thus obtained are found to agree well with those obtained from direct numerical simulations in the specific context of lattices with logistic map and henon map as on - site system dynamics .
we show how chimera states developed in an experimentally realizable 2-d lattice can be controlled using this scheme .
we propose this mechanism can provide a phenomenological model for the control of spatio - temporal patterns in coupled neurons due to non - synaptic coupling with the extra cellular medium .
we extend the control scheme to regulate dynamics on random networks and adapt the master stability function method to analyze the stability of the controlled state for various topologies and coupling strengths .
control of dynamics , amplitude death , chimera , coupled map lattice , random network 05.45.xt , 05.45.gg , 05.45.ra |
a linear differential system @xmath0 where @xmath1 are constant @xmath2-matrices , @xmath3 is diagonal , @xmath4 a parameter , is called an _ okubo system _ , or also a _
hypergeometric system_. such systems appear as a natural generalization of the hypergeometric equation .
it is known @xcite , that every single fuchsian differential equation can be reduced to such a system .
the assumption that @xmath3 is diagonal ( or semisimple ) assures that the 1-form @xmath5 has only simple poles ( placed at the eigenvalues of @xmath3 and at @xmath6 ) , i.e. that all the singularities of the okubo system are fuchsian .
the okubo system appears also as a dual to a system in _
birkhoff normal form _ @xmath7 which has an irregular singular point at 0 and a fuchsian singular point at @xmath6 , through the laplace
transform @xmath8 this fact can be used to express the stokes and connection matrices of the birkhoff system in terms of connection matrices and monodromies of the dual okubo system @xcite .
schfke @xcite has remarked that the the system can be also obtained from by the following confluence procedure : @xmath9 then @xmath10 satisfies @xmath11 which becomes at the limit when @xmath12 . in case of rank @xmath13 and @xmath3 with two distinct eigenvalues
, this confluence procedure corresponds exactly to the confluence of the ( gauss ) hypergeometric equation to the ( kummer s ) confluent hypergeometric equation . aside from the usual local levelt bases at each of the singularities , the okubo system has two other kinds of natural solution bases , so called _ mixed bases _ @xcite : the first one , called _ floquet basis _ , consists of the floquet solutions ( singular levelt solutions ) at different finite singularities @xmath14 ( eigenvalues of @xmath3 ) . the other , called _ co - floquet basis _
, is in a sense dual ; a co - floquet solution at a singularity @xmath14 is one that is analytic at al other singularities @xmath15 , @xmath16 .
schfke @xcite has studied the limits of these mixed bases in the confluent family in the case where all the eigenvalues of @xmath3 are distinct and has shown that they both tend to the canonical solution basis of the limit system ( borel sum of a formal fundamental solution ) on sectors at the irregular singularity @xmath17 : the floquet basis when @xmath18 , and the co - floquet basis when @xmath19 .
this article exposes these results while extending them to a more general situation , where @xmath3 is allowed to have multiple eigenvalues , and @xmath20 can go to infinity along any fixed direction in one of two sectors of opening @xmath21 covering a neighborhood of @xmath6 on the riemann sphere @xmath22 . in an analogy with @xcite
it is natural to introduce parametric _ stokes matrices _ of the confluent family , as connection matrices between different branches of the floquet ( resp .
co - floquet ) basis far from the origin .
these parametric stokes matrices are closely related to the monodromy of the family : in general , the monodromy matrices of the floquet and co - floquet bases can be expressed as products of these stokes matrices and formal monodromy matrices . while the monodromy matrices diverge when @xmath12 ( because of their formal monodromy parts which are exponential functions of @xmath20 ) these parametric stokes matrices tend to the usual stokes matrices of the limit system , and can be easily obtained from them ( proposition [ prop : hg - stokesmatrices ] ) .
these results are illustrated in section 2 on explicit calculations in the case of the generalized hypergeometric equation , previously studied by duval @xcite .
duval considered the problem of convergence of the monodromy matrices to the stokes matrices without separating formal monodromy part and the stokes part .
therefore she could only consider limits when @xmath23 following a discrete set of values on which the formal monodromy part is constant . a different confluence procedure of the type @xmath24 was investigated in e.g. @xcite . in case of @xmath3 having only two eigenvalues ( one of which can be always shifted to 0 ) , the confluence procedure can be considered as a special case of . in this case
our perspective essentially coincides with that of @xcite .
in particular , this includes the confluence in the gauss hypergeometric equations @xcite and in the generalized hypergeometric equation @xcite .
let the matrix @xmath3 be diagonal with eigenvalues @xmath25 of respective multiplicities @xmath26 , and let the matrix @xmath27 be partitioned into blocks accordingly @xmath28 the following * assumption * is made throughout the text : @xmath29 for any @xmath30-matrix @xmath31 , let @xmath32 be its bloc - partition according to @xmath3 , and denote @xmath33 the @xmath34-th bloc column of @xmath31 .
it is well - known ( see for example @xcite ) that the system can be bloc - diagonalized by means of a formal power series transformation @xmath35 , with @xmath36 under the assumption , the formally transformed system can be given the following birkhoff form @xmath37 therefore the system has a formal fundamental solution @xmath38 whose @xmath34-th bloc - column is given by @xmath39 while @xmath40 is in general divergent , it is borel summable .
more precisely each its column @xmath41 is borel summable in all directions @xmath42 with @xmath43 disjoint from all @xmath44 , @xmath45 ( such direction @xmath42 will be called _ non - singular _ ) .
let @xmath46 be the formal borel transform of @xmath47 , convergent near @xmath48 and extended analytically on the universal covering of @xmath49 .
the matrix function @xmath50 is a solution to linear system @xmath51 with fuchsian singularities at the points @xmath44 and @xmath6 . in particular
, @xmath52 has only a moderate growth at each of the singularities .
therefore the borel sum of @xmath41 in a non - singular direction @xmath42 is well - defined by the laplace integral @xmath53,\cdot j}(z)=\frac{1}{z}\int_0^{+\infty e^{i\alpha}}\!\!\ ! u_{\cdot j}(s)\,e^{-\frac{s}{z } } ds , \ ] ] which converges and is bounded for @xmath54 in the open half - plane bisected by @xmath43 , and whose value is independent of when the direction @xmath42 varies a bit . in another words , the sectoral transformation @xmath55}$ ]
depends only on the homotopy class @xmath56 $ ] of the direction @xmath57 , and one can consider it as defined on a sector in the @xmath54-plane @xmath58}(\infty):= \bigcup_{\alpha'\in[\alpha]}\{\re(e^{-i\alpha'}z)>0\},\ ] ] of opening @xmath21 .
once a branch of @xmath59 is fixed , the system has on each of these sectors a _ canonical solution basis _ @xmath60}(z)$ ] @xmath61,\cdot j}(z)=t_{[\alpha],\cdot j}(z)\ , z^{a_{jj } } e^{-\frac{\lambda_j}{z}}= \frac{1}{z}\int_0^{+\infty e^{i\alpha}}\!\!\ ! u_{\cdot j}(s)\,e^{-\frac{s}{z } } ds\cdot z^{a_{jj}}e^{-\frac{\lambda_j}{z}}.\ ] ] for every pair of non - singular directions @xmath62 there is _ stokes matrix _ @xmath63[\alpha_2]}(\infty)$ ] @xmath64}=\psi_{[\alpha_1]}\cdot s_{[\alpha_1][\alpha_2]}(\infty)\ ] ] ( defined by analytic continuation ) .
it is an easy fact that for two neighboring direction classes @xmath65,\ , [ \alpha_2]$ ] the stokes matrix @xmath63[\alpha_2]}(\infty)$ ] is unipotent with only non - zero off - diagonal entries at the positions @xmath66 corresponding to the singularity @xmath44 separating the direction classes @xmath65,\,[\alpha_2]$ ] .
note that by the liouville - ostrogradski formula @xmath67}$ ] is constant in @xmath54 and therefore equal to 1 .
the okubo system has @xmath68 fuchsian singularities on @xmath22 at the points @xmath14 , @xmath69 , and @xmath6 . near each @xmath70 ,
the system is written as @xmath71 @xmath72 denoting the @xmath34-th column bloc of the identity matrix , and @xmath73 standing for holomorphic terms that vanish at @xmath14 .
its local multipliers " are therefore @xmath74 in the @xmath34-th bloc and @xmath75 in the other @xmath76 blocs .
the system comes with two kinds of canonical mixed bases that will be of interest in this article .
the first one , which will be denoted @xmath77 , consists of the so called _ floquet solutions _
@xmath78 , which behave asymptotically like @xmath79 at the respective singularities @xmath14 .
the second one , denoted by @xmath80 , is in a sense dual to the floquet basis ; it consists of solutions @xmath81 that are analytic at each other singularity @xmath82 , @xmath45 .
this section describes these two bases in more detail .
let @xmath83 be a sector at @xmath6 in the parameter @xmath20-space , on which @xmath84\!-\!\pi+\eta,\pi-\eta[$ ] , with @xmath85 fixed arbitrary , and @xmath86 is sufficiently big so that @xmath87 .
symmetrically , let @xmath88 be a sector at @xmath6 on which @xmath84\eta,2\pi\!-\!\eta[$ ] , and @xmath86 is sufficiently big so that @xmath89 .
[ [ the - floquet - bases . ] ] the floquet bases .
+ + + + + + + + + + + + + + + + + + if @xmath90 and @xmath20 is large enough so that all eigenvalues of @xmath74 have positive real part , then the matrix function @xmath91 vanishes when @xmath92 approaches @xmath14 radially . correspondingly , consider the space of solutions of that vanish when @xmath93 radially
. can vanish at the singularity , they are asymptotically bigger and can not hide behind the vanishing solutions .
this is what makes this subspace of the space of solutions well - defined . ]
it is invariant by the local monodromy , and it follows from the local theory of fuchsian singularities ( cf .
@xcite ) and the assumption on @xmath94 , that this space has a unique basis written as @xmath95 this construction can be extended to all parameters @xmath96 , if instead of letting @xmath92 approach @xmath14 radially , one lets it approach @xmath14 following a suitable logarithmic spiral along which @xmath97 . more precisely , @xmath92 should follow a real positive trajectory of the vector field @xmath98 , for some @xmath99\!-\!\frac{\pi}{2},\frac{\pi}{2}[$ ] with @xmath100 .
the floquet solution @xmath101 is closely related to the @xmath34-th formal canonical solution of the dual birkhoff system . in fact , the formal borel transform ( = term - wise inverse laplace transform ) of @xmath102 equals to the convolution integral @xcite @xmath103 where the matricial gamma function is defined as usual by the integral @xmath104 and @xmath50 is given in .
the integral , also known as _ riemann - liouville integral _ with base - point at @xmath14 , solves , and moreover it satisfies @xmath105 , and therefore solves the difference equation @xmath106 the canonical solution @xmath60}$ ] of the birkhoff system equals @xmath107,\cdot j}(z)=z^{-\rho-1}\int_{\lambda_j}^{+\infty e^{i\alpha}}i_{\cdot j}^+(s,\rho)\,e^{-\frac{s}{z}}ds.\ ] ] the floquet solution @xmath101 is obtained from @xmath108 after a normalization : @xmath109 the integrating path in is such that @xmath110 follows a positive real trajectory of the vector field @xmath111 from the point @xmath14 to @xmath92 , with suitable @xmath112 as above , avoiding other singularities @xmath82 , @xmath45 , of @xmath113 .
the set of points @xmath92 which can be reached by such paths with varying @xmath112 then defines a ramified domain on which is defined
. note that if @xmath114 , then the integrating trajectory approaches @xmath14 in the asymptotic direction @xmath42 .
let @xmath115,j}(\rho)$ ] be a ( ramified ) domain consisting of those @xmath92 that can be reached by such trajectory for some direction @xmath42 in the given homotopy class , @xmath116,j}(\rho)\subseteq\{s\in\c\mid
\arg(s-\lambda_j)=\alpha'-\theta,\ and let @xmath115}(\rho):=\bigcap_j\omega^+_{[\alpha],j}(\rho)$ ] . the restriction of @xmath77 to @xmath115}$ ] will be denoted @xmath117}^+$ ] .
different homotopy classes of non - singular directions @xmath56 $ ] give rise to to different branches @xmath117}^+$ ] of @xmath77 near infinity .
[ [ the - co - floquet - bases . ] ] the co - floquet bases .
+ + + + + + + + + + + + + + + + + + + + + for given index @xmath34 , and a direction @xmath42 such that @xmath118 , @xmath45 , define the co - floquet solution @xmath117,\cdot j}^-$ ] at a singularity @xmath14 as the unique solution analytic on @xmath119 and having the following asymptotic behavior near @xmath14 : @xmath120,\cdot j}^-(s,\rho)=(e_j+{o}(s\!-\!\lambda_j))\cdot ( s\!-\!\lambda_j)^{a_{jj}+\rho},\ ] ] with @xmath121 denoting the usual landau symbol ( the corresponding terms may be ramified ) .
let s be more precise about where does it comes from .
for each singularity @xmath82 and @xmath122 large enough so that @xmath123 has no positive eigenvalue , define @xmath124 as the space of solutions analytic at @xmath82 .
it follows from the local theory of fuchsian singularities ( cf .
@xcite ) that this space is tangent exactly to the the @xmath76 vector - blocs @xmath125 , @xmath126 , corresponding to the multiplier 0 . for a point @xmath127 , continue each solution subspace @xmath124 , @xmath45 , toward @xmath92 in the cut plane @xmath119 , and define the subspace @xmath128 as their intersection .
since it consists of solutions analytic at each @xmath15 , @xmath16 , it does not depend on the way the @xmath129 are continued around the singularities @xmath15 , only on the direction @xmath42 of the cut .
following @xcite , there is a canonical bloc - solution of generating the space @xmath128 given by the integral : @xmath130}^-(s,\rho):=\int_{\lambda_j}^{+\infty e^{i\alpha } } u_{\cdot j}(\sigma\!-\!\lambda_j ) ( s\!-\!\sigma)^{a_{jj}+\rho-1}d\sigma \cdot\gamma(1\!-\!a_{jj}-\rho)e^{-\pi i ( a_{jj}+\rho)},\ ] ] which satisfies again @xmath131 , and therefore solves the difference equation @xmath132 the integral @xmath133}^-$ ] is a laplace transform of the canonical solution @xmath60}$ ] @xmath134}^-(s,\rho):=\int_{0}^{+\infty e^{i\alpha } } z^{\rho-1}\psi_{[\alpha],\cdot j}(z ) e^{\frac{s}{z } } dz,\ ] ] which in turn equals to @xmath135,\cdot j}(z)=z^{-\rho-1 } \frac{1}{2\pi i } \int_{\gamma_{j,\alpha } } i_{\cdot j}^-(s,\rho)\,e^{-\frac{s}{z}}ds,\ ] ] where the path @xmath136 encircles the ray @xmath137 in positive direction .
while @xmath60}$ ] is defined on a sector at 0 of an opening @xmath21 bisected by @xmath43 , the integral @xmath133}^-$ ] is defined on a sector at @xmath14 bisected by @xmath138 of an opening @xmath139 .
the co - floquet solution is obtained after a normalization @xmath140,\cdot j}^-(s,\rho)=\int_{\lambda_j}^{\infty } u_{\cdot j}(\sigma\!-\!\lambda_j ) ( s\!-\!\sigma)^{a_{jj}+\rho-1}d\sigma \cdot(a_{jj}\!+\!\rho).\ ] ] in the default situation when @xmath141 the integration path is the straight ray @xmath142 and the integral is defined for @xmath143 and extended analytically from there . in a general situation , the integration path follows a negative real trajectory of the vector field @xmath111 from the point @xmath14 to @xmath6 , with a suitable @xmath99\!-\!\frac{\pi}{2},\frac{\pi}{2}[$ ] such that @xmath144 , that is end - point homotopic to the ray @xmath137 in @xmath145 .
the set of points @xmath92 that can be reached by such paths defines again a ( ramified ) sectoral domain @xmath146,j}(\rho)\subseteq\{s\in\c\mid \arg(s-\lambda_j)=\alpha'-\theta+\pi,\ on which the integral is naturally defined .
let @xmath147}(\rho):=\bigcap_j\omega^-_{[\alpha],j}(\rho)$ ] .
[ prop : hg-1 ] for @xmath148 and @xmath149 , let @xmath150 , then @xmath151,\cdot j}^+(s,\rho)=\big[v_{[\alpha],\cdot j}^-(\tilde s,\rho)-v_{[\alpha],\cdot j}^-(s,\rho)\big]\cdot \big[e^{2\pi i(a_{jj}+\rho)}-1 \big]^{-1},\ ] ] or equivalently @xmath134,\cdot j}^+(s,\rho)=\frac{1}{2\pi i}\big[i_{[\alpha],\cdot j}^-(\tilde s,\rho)-i_{[\alpha],\cdot j}^-(s,\rho)\big],\ ] ] i.e. @xmath133,\cdot j}^+$ ] is a hyperfunction defined by the boundary value of @xmath152,\cdot j}^-$ ] on @xmath153
. follows from the construction .
the following proposition is due to okubo and kohno .
@xmath154}^+(s,\rho)&=\frac{\det \gamma(a_d+\rho+1)}{\det\gamma(a+\rho+1)}\cdot\det\,(s - b)^{a_d+\rho},\\[8pt ] \det v_{[\alpha]}^-(s,\rho)&=\frac{\det \gamma(-a-\rho)}{\det\gamma(-a_d-\rho)}\cdot\det\,(s - b)^{a_d+\rho}.\end{aligned}\ ] ] for the sake of completeness we will sketch here the proof in the co - floquet case ; the floquet case is almost identical and can be found in @xcite .
the co - floquet solution has the following asymptotic behavior w.r.t .
@xmath20 ( see @xcite , theorem ( 4.6 ) ) : @xmath155,\cdot j}^-(s,\rho)\cdot ( s\!-\!\lambda_j)^{-a_{jj}-\rho}=e_j+{o}(\frac{1}{|\rho|}),\quad\text{when } \
\re(\rho)\to -\infty,\ ] ] locally uniformly in the cut plane @xmath156 .
now , for @xmath157 it follows from by induction that @xmath134,\cdot j}^-(s,\rho\!-\!m)\cdot ( s\!-\!\lambda_j)^{m}= ( a\!+\!\rho\!-\!m\!+\!1)\cdots(a\!+\!\rho)\cdot i_{[\alpha],\cdot j}^-(s,\rho),\ ] ] and hence @xmath158,\cdot j}^-(s,\rho - m)\cdot ( s\!-\!\lambda_j)^{-a_{jj}-\rho+m}=\\ & \qquad = v_{[\alpha],\cdot j}^-(s,\rho ) \cdot ( s\!-\!\lambda_j)^{-a_{jj}-\rho}\cdot \gamma(-a_{jj}\!-\!\rho)\cdot\gamma(-a_{jj}\!-\!\rho+\!m)^{-1}.\end{aligned}\ ] ] therefore @xmath159}^-(s,\rho - m)\cdot ( s\!-\!b)^{-a_{d}-\rho+m}\right]=\\ & = \det\left[v_{[\alpha]}^-(s,\rho)\cdot ( s\!-\!b)^{-a_{d}-\rho}\right]\cdot \frac{\det\gamma(-a_d\!-\!\rho)}{\det\gamma(-a\!-\!\rho)}.\end{aligned}\ ] ] letting @xmath160 and using and usual formulas for the @xmath161-function , one can see that both expressions on the left side tend to 1 . for @xmath96 ( resp .
@xmath122 ) the floquet ( resp .
the co - floquet ) solutions form a basis of the solution space .
the system has two kinds of canonical fundamental matrix solutions @xmath162}^\pm(z,\rho)$ ] corresponding to the floquet and co - floquet bases of .
in order to obtain a convergence when @xmath12 , one has to be a bit careful with the choice of their branch .
it is convenient to write them as @xmath163}^\pm(z,\rho)=t_{[\alpha]}^\pm(z,\rho)\cdot \phi(z,\rho),\ ] ] where @xmath164 is a canonical solution to the bloc - diagonal system @xmath165 whose branch needs to be selected so that it converges to the adequate branch of @xmath166 when @xmath12 .
the bloc - diagonalizing transformation @xmath55}^\pm$ ] is defined by @xmath53}^\pm(z,\rho)=v_{[\alpha]}^\pm(\rho z,\rho)\cdot(\rho z - b)^{-a_{d}-\rho},\ ] ] where the branch of @xmath167 is chosen in accord with the one inside the integral , . hence @xmath168,\cdot
j}^+(z,\rho)=\int_0^{\rho z-\lambda_j } u_{\cdot j}(\sigma ) \big(\frac{\rho
z-\lambda_j-\sigma}{\rho z-\lambda_j}\big)^{a_{jj}+\rho}\frac{d\sigma}{\rho z-\lambda_j-\sigma } \cdot(a_{jj}+\rho),\ ] ] where the integration path follows a positive trajectory of the vector field @xmath169 from the point 0 to @xmath170 , and the branch of @xmath171 is chosen so that it is equal 1 at the endpoint .
remark , that at the limit , when @xmath12 radially with fixed @xmath172 , the trajectories of the given vector field become trajectories of the vector field @xmath173 with @xmath174 .
therefore the integral has a well - defined limit @xmath53,\cdot j}^+(z,\infty)=\frac{1}{z}\int_0^{+\infty e^{i\alpha } } u_{\cdot j}(\sigma)\ , e^{-\frac{\sigma}{z } } d\sigma.\ ] ] similarly , @xmath175,\cdot j}^-(z,\rho)=\int_0^{\infty } u_{\cdot j}(\sigma ) \big(\frac{\rho
z-\lambda_j-\sigma}{\rho z-\lambda_j}\big)^{a_{jj}+\rho}\frac{d\sigma}{\rho z-\lambda_j-\sigma } \cdot(a_{jj}+\rho),\ ] ] where the integration path follows a positive trajectory of @xmath176 from the point 0 to @xmath6 , which at the limit , when @xmath12 radially , becomes a trajectory of @xmath177 with @xmath178 , and the integral becomes @xmath53,\cdot j}^-(z,\infty)=\frac{1}{z}\int_0^{+\infty e^{i\alpha } } u_{\cdot j}(\sigma)\ , e^{-\frac{\sigma}{z } } d\sigma.\ ] ] the transformations @xmath55}^\pm(\cdot,\rho)$ ] are defined on sectors @xmath179}^\pm(\rho):=\tfrac{1}{\rho}\omega^\pm_{[\alpha]}(\rho),\ ] ] which tend to a subsector of @xmath180}(\infty)$ ] depending on the radial direction in which @xmath12 .
[ [ stokes - matrices - of - the - confluent - family ] ] stokes matrices of the confluent family + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + fixing a branch of @xmath181 near @xmath182 and its restrictions to the sectors @xmath180}^\pm(\rho)$ ] one obtains a canonical set of fundamental matrix solutions @xmath162}^\pm(z,\rho)$ ] .
the connection matrices between these solutions near @xmath182 corresponding to different non - singular directions @xmath62 @xmath183}^\pm = y_{[\alpha_1]}^\pm\cdot s_{[\alpha_1][\alpha_2]}^\pm(\rho)\ ] ] will be called _ stokes matrices _ of the family . [
prop : hg - stokesmatrices ] @xmath184[\alpha_2]}^+(\rho)&=\rho^{a_d}\gamma(a_d+\rho+1)^{-1 } s_{[\alpha_1][\alpha_2]}(\infty)\,\gamma(a_d+\rho+1)\rho^{-a_d } , \\
s_{[\alpha_1][\alpha_2]}^-(\rho)&=\big(e^{-\pi i}\rho\big)^{-a_d}\gamma(-a_d-\rho)\ , s_{[\alpha_1][\alpha_2]}(\infty)\,\gamma(-a_d-\rho)^{-1}\big(e^{-\pi i}\rho\big)^{-a_d},\end{aligned}\ ] ] which tends to the stokes matrix @xmath63[\alpha_2]}(\infty)$ ] when @xmath12 in @xmath185 respectively . for two neighboring direction classes @xmath65,\ ,
[ \alpha_2]$ ] the stokes matrix @xmath63[\alpha_2]}^\pm(\rho)$ ] is unipotent with only non - zero off - diagonal blocs at the positions @xmath186 corresponding to the direction of @xmath187 separating @xmath188 .
this follows from the relation @xmath189}^\pm(s,\rho)=i_{[\alpha_1]}^\pm(s,\rho)\cdot s_{[\alpha_1][\alpha_2]}^\pm(\rho),\ ] ] which is a consequence of the formulas , resp . .
0.05 [ example : hg-1 ] suppose @xmath3 has just three eigenvalues @xmath82 , @xmath190 and assume they are not colinear . for simplicity we will consider only the default situation when @xmath191 and restrict the domains of @xmath162}^\pm$ ] to a smaller sector consisting of the points @xmath192 for which the integration path in , resp .
, can be taken a straight segment .
near @xmath6 , these sectors are are separated by the outer parts of lines through @xmath193 , whose crossing is governed by the stokes matrices . for each singularity @xmath194 or @xmath6
make a cut ( wavy line in figure [ figure : hg-1 ] ) from the origin on which the formal solution @xmath195 is branched , and therefore changed by its formal monodromy @xmath196 & \ddots & & & \\[1pt ] & & \hskip-12pt e^{2\pi i(a_{jj}+\rho)}\hskip-18pt & & \\[-4pt ] & & & \ddots & \\ & & & & i_{n_3}\end{smallmatrix}\right),\quad j=1,2,3 , \quad\text{and}\quad n_\infty = e^{-2\pi i a_{d}}.\ ] ]
this section illustrates the confluence procedure on the example of the generalized hypergeometric equation , where things can be expressed very explicitly .
most of the formulas come from @xcite . to simplify the writing we adopt the following notation . in the above notation , _ the generalized hypergeometric equation _ of order @xmath206 , is written as @xmath207 , where @xmath208 and @xmath209 is the euler operator .
it has three regular singular points at @xmath75 , @xmath210 and @xmath6 .
since @xmath211 one can always use the transformation @xmath212 , to bring the equation to a more usual form in which one of the @xmath213 s equals 1 : @xmath214 .
the equation has thus @xmath206 local solution at @xmath48 given by the hypergeometric series @xmath215 convergent for @xmath216 , where @xmath217 denotes the pochhammer symbol @xmath218 these solutions are linearly independent if no two @xmath213 s differ by an integer .
we are interested in the situation when @xmath223 .
the situation when @xmath224 would be similar due to the symmetry .
let @xmath225 then @xmath226 where the regular singularities at @xmath17 and @xmath227 merge for @xmath12 to form an irregular singularity .
setting @xmath228 , with @xmath229 the equation : @xmath230 is written in the form of a family of systems with @xmath231 & \ddots & & \\ & & 0 & \\ & & & 1 \end{pmatrix},\qquad a=\begin{pmatrix } 1\!-\!\beta_1\hskip-9pt & \hskip6pt 1\hskip-6pt & & \\[-3pt ] & \ddots & \ddots & \\ & & \hskip-12pt 1\!-\!\beta_{n-1 } \hskip-6pt & 1 \\ * & \ldots & * & \gamma \end{pmatrix},\ ] ]
@xmath232 where @xmath27 has @xmath233 as eigenvalues . for @xmath238
, we have @xmath239 singular solutions of the okubo system near @xmath48 whose first component is given by @xmath240 and one singular solution at @xmath241 given by meijer g - function @xmath242 with @xmath243 and the coefficients @xmath244 independent of @xmath20 ( see @xcite , p. 601 ) @xmath245 it is easy to see that the terms of the fundamental solution matrix @xmath246 have the asymptotic behavior @xmath247 where the upper - triangular matrix @xmath248 @xmath249 commutes with @xmath3 and diagonalizes @xmath250 & \raisebox{0.15\normalbaselineskip}{$\ddots$ } & \hskip-3pt\raisebox{0.7\normalbaselineskip}{$\ddots$ } & \hskip-12pt 1 & \\ & & & \hskip-24pt 1\!-\!\beta_{n-1 } \hskip-6pt & \\ & & & & \,\gamma \end{pmatrix } , \qquad r^{-1}\!a_dr=\tilde a_d:=\begin{pmatrix } 1\!-\!\beta_1\hskip-9pt & \hskip3pt & & \\[-3pt ] & \ddots & & \\ & & \hskip-12pt 1\!-\!\beta_{n-1 } \hskip-6pt & \\ & & & \,\gamma \end{pmatrix}.\ ] ] therefore , @xmath251 is the floquet bases of the okubo system with the asymptotic behavior , while @xmath252 is the floquet basis of the okubo system with @xmath253 , @xmath254 .
monodromy matrices @xmath256 and @xmath257 of the solution @xmath258 around the singularities @xmath75 and @xmath210 in the positive direction from a base - point at @xmath259 are calculated in @xcite : @xmath260 & \hskip-6pt\ddots & & \vdots \\ & & \hskip-6pt e_{n-1 } \hskip-3pt & \xi_{n-1}(e_{n-1}\!-\!1)\hskip-3pt\\[3pt ] & & & 1 \end{pmatrix } , \qquad \tilde m_1^+= \begin{pmatrix } 1 & & & \\[-3pt ] & \ddots & & \\ & & \hskip-6pt 1 & \\[3pt ] \eta_1(e_n\!-\!1 ) \hskip-4pt & \ldots & \hskip-4pt \eta_{n-1}(e_n\!-\!1)\hskip-4pt & e_n \end{pmatrix},\ ] ] where @xmath261 and @xmath262 the connection matrix between the floquet and the co - floquet bases @xmath263 is also calculated in @xcite : @xmath264 & \ddots & & \vdots\\ & & \hskip-6pt 1 & \hskip-6pt -\xi_{n-1}\hskip-3pt\\[3pt ] -\eta_1 \hskip-3pt & \ldots & \hskip-3pt -\eta_{n-1 } \hskip-6pt & 1 \end{pmatrix}.\ ] ] therefore the corresponding monodromy matrices of @xmath265 are equal to @xmath266 @xmath267 & \hskip-6pt\ddots & & \\ & & \hskip-20pt e_{n-1 } & \\[3pt ] \eta_1(e_1\!-\!1 ) \hskip-4pt & \ldots & \hskip-4pt \eta_{n-1}(e_{n-1}\!-\!1)\hskip-4pt & 1 \end{pmatrix } , \qquad \tilde m_1 ^ -= \begin{pmatrix } 1 & & & \xi_1(e_n\!-\!1 ) \\[-3pt ] & \ddots & & \vdots\\ & & 1 & \xi_{n-1}(e_n\!-\!1)\hskip-3pt\\[3pt ] & & & e_n \end{pmatrix}.\ ] ] under the assumption there are @xmath270 parameter - dependent singular solutions of the confluent equation near @xmath17 are given for @xmath238 by @xmath271 with the limit asymptotic on compact sub - sectors to the divergent formal series @xmath272 and the singular solution at @xmath227 is is given by @xmath273 which is asymptotic on compact sub - sectors to the divergent formal series @xmath274 the constructed fundamental matrix solution @xmath275 have the asymptotic behavior @xmath276 therefore @xmath277 is the corresponding floquet bases of the confluent family with the right asymptotic behavior . in the definition of @xmath278 above the right choice of branch of @xmath59 and @xmath279 in @xmath280
is of essential importance .
let @xmath42 be a direction , @xmath281 , and chose the bases @xmath282}$ ] and @xmath283}$ ] so that they are related to each other as in figure [ figure : hg - y ] , @xmath284}^+(z,\rho)=(\rho z)^{-\rho}\tilde v^+(\rho z , \rho)\rho^{-\tilde
a_d},\quad \tilde y_{[\alpha]}^-(z,\rho)=(\rho z)^{-\rho}\tilde v^-(\rho z , \rho)\rho^{-\tilde a_d}.\ ] ] } $ ] , @xmath283}$ ] on their natural domains ( restricted to a fixed neighborhood of 0 ) and their transition matrices , according to the values of @xmath255 .
the limit system is in the center .
( see @xcite for more details on the construction of these ramified domains . ) ] then the monodromy matrices of the fundamental matrix solution @xmath285}^+(z,\rho)$ ] , resp .
@xmath286}^-(z,\rho)$ ] , around @xmath75 and @xmath255 ( @xmath238 ) in the positive direction from a base - point at @xmath287 are equal @xmath288 resp .
@xmath289 where @xmath290 & \hskip-6pt\ddots & & \\ & & e^{2\pi i(1-\beta_{n-1 } ) } \hskip-24pt & \\ & & & e^{-2\pi i\rho } \end{pmatrix } , \qquad \tilde n_{\frac{1}{\rho}}= \begin{pmatrix } 1 & & & \\[-3pt ] & \ddots & & \\ & & 1 & \\ & & & e^{2\pi i(\gamma+\rho ) } \end{pmatrix},\ ] ] and @xmath291 are monodromies of the fundamental matrix solution @xmath292 of the diagonal model system , and @xmath293 & \hskip-3pt\ddots\hskip-3pt & & \hskip3pt\vdots\ \\ & & 1 \hskip-6pt & \tilde s^\pm_{n-1,n}\hskip-3pt \\[3pt ] & & & 1 \end{pmatrix } , \qquad \tilde s_l^\pm:=\tilde s_{\alpha+\pi,\alpha}^\pm= \begin{pmatrix } 1 & & & \\[-3pt ] & \ddots & & \\ & & \hskip-6pt 1 & \\[3pt ] \tilde s^\pm_{n1 } \hskip-6pt & \ldots & \hskip-6pt\tilde s^\pm_{n , n-1}\hskip-6pt & 1 \end{pmatrix}\ ] ] are the stokes matrices .
it follows from that the stokes multipliers @xmath294 are equal to : @xmath295 @xmath296 and @xmath297 from one then obtains the stokes multipliers @xmath298 : @xmath299 @xmath300 , and @xmath301 one could also proceed the opposite way : the stokes matrices @xmath302 , @xmath303 , of the limit generalized confluent hypergeometric equation have been calculated in @xcite , and the stokes matrices @xmath304 of the family are related to them via proposition [ prop : hg - stokesmatrices ] .
note that the confluent floquet and co - floquet bases @xmath305 and their monodromies , resp .
stokes matrices @xmath306 , @xmath307 , resp .
@xmath308 , @xmath303 , are well - defined under a weaker assumption than , that no two @xmath213 s differ by a non - zero integer .
w. balser , w.b .
jurkat , d.a .
lutz , _ on the reduction of connection problems with an irregular singularity to ones with only regular singularities , i , ii _ , siam j. math
* 12 * ( 1981 ) , 691721 , siam j. math . anal . * 19 * ( 1988 ) , 398443 .
j. hurtubise , c. lambert , c. rousseau , _ complete system of analytic invariants for unfolded differential linear systems with an irregular singularity of poincar rank k _ , moscow math .
j. * 14 * ( 2013 ) , 309338 .
l. parise , _ confluence de singularits rgulires dquations diffrentielles en une singularit irrgulire . modle de garnier _ , thse de doctorat , irma strasbourg ( 2001 ) .
[ http://www-irma.u-strasbg.fr/annexes/publications/pdf/01020.pdf ] | a system in a birkhoff normal form with an irregular singularity of poincar rank 1 at the origin and a regular singularity at infinity is through the borel - laplace transform dual to a system in an okubo form .
schfke has showed that the birkhoff system can also be obtained from the okubo system by a simple limiting procedure .
the okubo system comes naturally with two kinds of mixed solution bases , both of which converge under the limit procedure to the canonical solutions of the limit birkhoff system on sectors near the irregular singularity at the origin .
one can then define stokes matrices of the okubo system as connection matrices between different branches of the mixed solution bases and use them to relate the monodromy matrices of the okubo system to the usual stokes matrices of the limit system at the irregular singularity .
this is illustrated on the example of confluence in the generalized hypergeometric equation . |
low dimensional organic conductors have attracted sustained interest regarding their rich phase diagram showing almost all known ground states with particularly a superconducting phase next to a magnetic ordering @xcite , which is reminiscent of high-@xmath0 superconductors .
this proximity is believed to result in a phase segregation consisting of an inhomogeneous coexistence of electronic states such as the nanoscale charge stripes observed in copper oxide @xcite .
mesoscopic scale inhomogeneities have been reported in the organic charge transfer salts ( tmtsf)@xmath2x @xcite , so - called bechgaard salts , and @xmath3(bedt - ttf)@xmath2y @xcite where tmtsf denotes the donor molecule tetramethyl - tetraselenafulvalene while bedt - ttf stands for the bis(ethylenedithio)tetrathiafulvalene ; x = clo@xmath4 , pf@xmath5 , reo@xmath4 ... and y = cu(ncs)@xmath2 , cu[n(cn)@xmath2]z ( z = br and cl ) are inorganic anions .
these compounds are highly anisotropic : the former show marked quasi - one dimensional ( q1d ) electronic properties whereas the latter , abbreviated as ( et)@xmath2 , can be regarded as a layered ( two - dimensional 2d ) systems .
the switching between the several low temperature electronic phases of these compounds can be controlled by applying hydrostatic pressure or by chemical substitution of either the inorganic anion or the donor molecule . a close investigation of the boundary between the magnetic phase and the superconducting one reveals a mesoscopic phase segregation where the system shows coexistence of superconducting domains separated by magnetic ones @xcite .
several experimental studies have reported that the interplay between these two phases is substantially dependent on the purity of the sample .
likewise , it has been found that this segregated structure can be induced by varying the hydrostatic pressure around the critical value @xmath7 at which the magnetic phase collapses @xcite .
disorder turns out to be a key parameter for the stability of the superconducting state in these compounds .
despite the general consensus on the unconventional nature of the superconducting phase of different organic superconductors , the symmetry of the corresponding order parameter is still under debate @xcite . studying the effect of disorder on the superconducting phase
may bring an answer to the puzzling question of the superconducting symmetry . in these molecular materials ,
two types of disorder can be found : ( i ) extrinsic disorder resulting from the random imperfections obtained by introducing impurities or radiation damages .
( ii ) intrinsic disorder related to the noncentrosymmetric character of the inorganic anions , such as clo@xmath8 in ( tmtsf)@xmath2clo@xmath4@xcite , or to the presence of internal degrees of freedom such as partially disordered ethylene groups adopting several conformations in @xmath3(et)@xmath2x @xcite .
these structural degrees of freedom , which are strongly dependent on the cooling history of the sample , have drastic effect on the superconducting state .
a rapid cooling of the sample induces a strong depression of the superconducting transition temperature @xmath0 and may even destroy superconductivity as in ( tmtsf)@xmath2clo@xmath4@xcite .
there have been several experimental studies dealing with the disorder in organic superconductors . in ( tmtsf)@xmath2x , studies were focused particularly on the cooling rate induced disorder in ( tmtsf)@xmath2clo@xmath4 @xcite and disorder generated by the anion substitution in ( tmtsf)@xmath2(clo@xmath4)@xmath9(reo@xmath4)@xmath10 @xcite . in @xmath3(et)@xmath2x , disorder was introduced by rapid cooling @xcite , chemical substitution of donor or anion @xcite , x - ray and proton damage @xcite . the superconducting transition temperature t@xmath11 is found to be reduced by increasing the amount of disorder regardless of its origin @xcite . a theoretical analysis @xcite of the dependence of t@xmath11 on disorder in different ( et)@xmath2x salts has shown that the suppression of t@xmath11 obeys to the abrikozov - gorkov ( ag ) law @xcite which describes the suppression of t@xmath11 in the presence of either magnetic impurities or non magnetic impurities in non - s - wave superconductors . however , for large disorder , clear departure from the ag formula has been reported in several low dimensional organic superconductors @xcite .
alternative mechanisms have been suggested to account for this peculiar behavior @xcite .
a more detailed discussion is given below .
the outcomes of the studies dealing with the effect of disorder on organic superconductors are consistent with the unconventional character of the superconducting state since the latter is suppressed in presence of non - magnetic impurities@xcite . however , these studies do not state clearly the symmetry of the superconducting order parameter .
experiments carried out on bechgaard salts @xcite show that the upper critical fields @xmath1 along the most conducting axes @xmath12 and @xmath13 , @xmath14 and @xmath15 , are clearly times larger than the pauli paramagnetic limit @xmath16 above which @xmath14 and @xmath15 continue to increase with no sign to saturation . this behavior seems to rule out the singlet state and gives compelling indications of the presence of a spin triplet superconducting phase .
however , the question of the superconducting gap symmetry in bechgaard salts is not settled yet . despite the general tendency to accept the singlet character of the low field state
, the symmetry of the high field phase continues to be a controversial issue @xcite . in this paper , we first complete the theoretical study presented in ref.@xcite dealing with the role of cooling induced disorder on the inhomogeneous superconducting phase of ( tmtsf)@xmath2clo@xmath4 where it is found that t@xmath11 is strongly suppressed as the mixed character is enhanced .
we then present a quantitative comparison with experiments which was lacking in ref.@xcite . in the present work ,
we give a detailed analysis of the experimental data in quasi-1d and 2d organic superconductors to infer the key parameters used in our model .
we then derive the behavior of t@xmath11 as a function of various types of disorder . to the best of our knowledge
our results are the first to account quantitatively for the suppression of the superconducting transition temperature by various sources of disorder in several organic superconductors .
there is a general consensus , emerging from recent experiments , on the discrepancy between the experimental data obtained at large disorder rate and the results expected from the ag law @xcite .
our work will shed light on the origin of this discrepancy .
we also briefly present in this paper , the effect of inhomogeneity on the upper critical fields in the case of layered organic superconductors where superconducting domains are sandwiched between non - superconducting regions as it has been reported in ( tmtsf)@xmath2pf@xmath5 @xcite . in the next section , we summarize the experimental results related to the dependence of t@xmath11 on disorder in several low dimensional organic superconductors . after a brief comment on the theoretical results presented in literature , we describe , in section 3 ,
the method we used to extract , from the experimental data , the theoretical parameters of the model we described in ref.@xcite . in section 4 , results are discussed in connection with experiments .
it is worth noting that , as done in refs.@xcite , we do not consider a particular symmetry of the superconducting gap .
however , the decrease of t@xmath11 as a function of non magneticc disorder , reported in the present paper , supports the scenario of unconventional superconducting order parameter in the different low dimensional organic superconductors discussed in this work .
this scenario seems to be the subject of a general consensus emerging from the huge experimental results dealing with these compounds @xcite .
nevertheless , the debate on the gap symmetry is not completely settled .
when cooled sufficiently slowly , the clo@xmath8 anions in ( tmtsf)@xmath2clo@xmath4 order for entropy reasons @xcite at a structural transition temperature @xmath17 k. the nature of the electronic ground state , stabilized below @xmath18 , is found to be substantially dependent on the cooling rate . in the relaxed state ( r - state ) , obtained for a cooling rate less than 0.1k / min , the ground state is superconducting below 1.2 k @xcite .
however , the quenched state ( q - state ) resulting of a cooling at a rate more than 50k / min , is an insulating spin density wave ( sdw ) @xcite .
the most striking effect of the cooling rate is obtained for intermediate cooling rates at which the sample exhibits an inhomogeneous mixture of superconducting domains ( in which clo@xmath4 anions are ordered ) and sdw regions ( where clo@xmath4 anions are disordered)@xcite.meissner effect signal @xcite showed that , by increasing the cooling rate , the insulating regions get larger at the expense of the superconducting ones .
these results have been corroborated by x - ray study @xcite showing a variation of the fraction of ordered clo@xmath4 domains as a function of the cooling rate .
the fraction value was found to be identical to that of the superconducting volume deduced from the meissner measurement @xcite .
the x - ray study has also provided the average size of these ordered regions as a function of the cooling rate . +
this phase segregation has been also induced by x - ray irradiation @xcite and chemical substitution@xcite in ( tmtsf)@xmath2clo@xmath4 and by hydrostatic pressure in ( tmtsf)@xmath2pf@xmath19 @xcite .
it is worth to stress that the possibility of a sdw / superconducting phase coexistence in ( tmtsf)@xmath2pf@xmath5 was first suggested by greene _
@xcite who also considered a filamentary superconducting domain structure .
+ the signature of such coexistence has been also reported in the organic superconductor ( mdt - ts)(aui@xmath2)@xmath20 , where mdt - ts denotes methylenediothio - tetraselenafulvalene , showing a temperature - pressure phase diagram reminiscent of those obtained in ( tmtsf)@xmath2x and @xmath3(et)@xmath2x salts @xcite .
+ several experimental studies have revealed that the cooling rate has a drastic effect on the low temperature electronic properties of @xmath3(et)@xmath2x ( x = cu[n(cn)@xmath2]br , cu[n(cn)@xmath2]cl and cu(ncs)@xmath2 ) @xcite .
different studies of the hydrogenated @xmath3(et)@xmath2cu[n(cn)@xmath2]br and the deuterated analog @xmath3(d@xmath21-et)@xmath2cu[n(cn)@xmath2]br have given evidence of inhomogeneous superconducting state whose volume fraction is found to be reduced by a quenching the sample @xcite .
such inhomogeneity may explain the anomalous behavior of the inplane conductivity in these materials @xcite .
it has been also found that the inhomogeneous superconductivity in @xmath3(et)@xmath2x compounds can also be generated by chemical @xcite or hydrostatic pressure @xcite and irradiation @xcite .
+ the outcome of the experimental studies on @xmath3(et)@xmath2x and ( tmtsf)@xmath2x materials is that the proximity of the superconducting state to a magnetic insulating phase is a key ingredient for the formation of an inhomogeneous phase . in the following , we present a theoretical approach to account for the dependence of the superconducting transition temperature on the disorder amount regardless of its origin .
we assume that the disorder ds not introduce local magnetic impurities in the ordered domains which is consistent with experimental findings .
in this section , we first give a brief review of the theoretical approaches proposed in the literature to explain the behavior of t@xmath11 as a function of disorder in organic conductors .
we then present the outlines of the model we derived in ref.@xcite to qualitatively interpret the dependence of t@xmath11 on the cooling rate in ( tmtsf)@xmath2clo@xmath4 .
the key point lacking in ref.@xcite is the correspondence between the theoretical parameters and the experimental disorder texture necessary to provide a quantitative comparison with the experimental results . in the remainder of the section ,
we propose a method the extract the values of these parameters from the experimental data .
powell and mckenzie @xcite have analyzed , in the framework of the ag law , the experimental results dealing with the effect of different types of disorder on the superconducting transition temperature t@xmath11 of quasi-2d organic conductors .
it is worthwhile reminding that the ag law describes the suppression of t@xmath11 by magnetic impurities for s - wave pairing . for
non - s - wave superconductors , t@xmath11 is suppressed by non - magnetic impurities according also to the ag formula given by @xcite : @xmath22 where @xmath23 is the superconducting critical temperature in the pure limit and @xmath24 is the digamma function . @xmath25 is the quasi - particle lifetime due to scattering with magnetic impurities ( @xmath26 ) or non - magnetic impurities ( @xmath27 ) .
powell and mckenzie @xcite compared the experimental data obtained in @xmath28(et)@xmath2x and @xmath3(et)@xmath2x with various types of disorder to the expectations of the ag formula .
they found an excellent agreement using two free parameters to fit the theory .
however , their study was not conclusive concerning the conventional nature of the superconducting order parameter , since the agreement with ag law holds for both magnetic and non - magnetic impurities .
we emphasize that the data analyzed by powell and mckenzie are in the clean limit corresponding to small amounts of disorder which acts as scattering points .
it is worth noting that joo _
_ @xcite have also reported that the suppression of t@xmath11 with increasing impurity concentration @xmath29 in the solid solution ( tmtsf)@xmath2(clo@xmath4)@xmath9(reo@xmath4)@xmath10 is consistent with ag formula . moreover ,
based on epr measurements , the authors @xcite ruled out the magnetic nature of the scattering centers .
the sensitivity of t@xmath11 to the impurity concentration was , therefore , taken as a signature of the unconventional character of the superconducting gap in ( tmtsf)@xmath2clo@xmath4 . despite the good agreement between the ag law and the experimental results in the clean limit of organic conductors ,
a clear discrepancy has been reported in the dirty limit of quasi-2d salts @xcite . to account for this discrepancy , analytis _ et al._@xcite suggested to use a generalized ag formula involving two order parameters with both s - wave and unconventional components .
the presence of two gradients in the impurity induced suppression of t@xmath11 was ascribed to this mixed superconducting ordering .
the rapid decrease of t@xmath11 for small disorder amount is attributed to the non - s - wave part whereas the slowing down of the t@xmath11 suppression is assigned to the s - wave component .
however , to the best of our knowledge , there is no experimental evidence of such mixed superconducting gap in organic conductors @xcite .
+ recently , we have proposed a model to describe the suppression of the superconducting transition temperature with increasing cooling rate in ( tmtsf)@xmath2clo@xmath4 @xcite .
basically , the model concerns a layered superconductor , where each layer , corresponding to the most conducting plane , has a phase segregation structure with superconducting domains embedded in a non - superconducting host . for simplicity , we assume that the superconducting clusters are identical and form a square lattice . these superconducting regions are interacting via josephson couplings parameterized by @xmath30 and @xmath31 along the inplane directions @xmath12 and @xmath32 respectively .
these domains are also coupled along the transverse direction @xmath33 by the interplane josephson coupling @xmath34 .
this structure is consistent with the one proposed by mller _
@xcite from fluctuation spectroscopy measurements in @xmath3(et)@xmath2cu[n(cn)@xmath2]cl .
the authors have also argued that the mixed state of superconducting and non - superconducting regions can be regarded as a network of josephson coupled junctions connecting the superconducting clusters .
however , one should emphasize that in real samples the superconducting domains have not necessary planar structure with identical shape and size .
for example , x ray measurements in ( tmtsf)@xmath2clo@xmath4 @xcite suggested ellipsoidal structure .
one should also consider randomly distributed superconducting clusters instead of the square lattice assumed in our model
. actually , taking into account the thickness of the superconducting domain along the least conducting axis will not change the overall behavior of our results since superconductivity in the considered materials has essentially a two - dimension ( 2d ) character .
moreover , the shape of the superconducting domains is not a key factor in our model as far as the corresponding size is larger than the coherence length to avoid the phase fluctuations complexity .
the basic parameter is the size of the non - superconducting junction on which depend the josephson couplings
. it should be interesting to study the effect of a random distribution of the superconducting clusters which may be introduced by assuming a spatial weight dependence of the inplane josephson coupling parameters .
however , such correction is not expected to affect the outcomes of the present model .
indeed , the results depend on parameters extracted from experimental data which can be regarded as mean values associated to the randomly distributed superconducting islands in the real material .
it turns out , that considering different sized superconducting clusters with a random distribution may improve the quantitative agreement between our results and experiments but does not affect the general behavior .
in ( tmtsf)@xmath2clo@xmath4 and @xmath3(et)@xmath2x salts , the josephson junction is basically antiferromagnetic .
josephson effects through magnetic junctions have been discussed by andersen _
the anisotropic structure of the organic superconductors is at the origin of strong superconducting fluctuations which are considerably enhanced as the dimensionality of the system is reduced @xcite .
superconducting fluctuations , which can be brought out in many physical properties , express the presence of superconducting precursor effects in the normal state far from the superconducting transition temperature .
these fluctuations smear out the sharp superconducting phase transition anomaly observed in 3d isotropic superconductors .
in anisotropic systems , this transition is broadened and spread out over a transient regime where superconducting fluctuations take place @xcite .
several experimental studies have given evidence for the presence of superconducting fluctuations in organic conductors @xcite .
the critical superconducting fluctuations occur in the vicinity of @xmath0 over a thermal interval @xmath35 characterized by large fluctuations of the order parameter .
the strength of the critical fluctuations is given by the so - called ginzburg number @xcite : @xmath36 ^ 2\end{aligned}\ ] ] where the anisotropic parameter @xmath37 is the ratio of the inplane coherence length over the out - of - plane one , @xmath38 is the zero temperature inplane penetration depth and @xmath39 is the quantum flux . for conventional superconductors
@xmath40 is about @xmath41 @xcite whereas for hight-@xmath0 cuprates , @xmath40 amounts to @xmath42 @xcite . for ( et)@xmath2x and ( tmtsf)@xmath2x salts ,
@xmath40 is of the order of @xmath43 @xcite . due to the superconducting fluctuations
, t@xmath11 is reduced compared to the mean - field value @xmath44 obtained in a bulky superconductor in the absence of fluctuations @xcite .
@xmath44 can be estimated as @xmath45 where @xmath46 is the experimental critical temperature obtained in the pure limit .
the critical dynamics of the superconducting order parameter is governed by the time dependent ginzburg - landau ( tdgl ) equation @xcite @xmath47 where @xmath48 is the superconducting order parameter in the ( @xmath49 ) domain of the @xmath50 plane . here
@xmath51 ( @xmath52 ) labels the superconducting island along the @xmath12 ( @xmath32 ) direction .
@xmath53 is the relaxation rate of the order parameter , @xmath54 is the inplane coherence length and @xmath55 is the effective pair mass in the ( @xmath56 ) plane .
the langevin forces @xmath57 in eq.[partial ] govern the thermodynamic fluctuations of the superconducting order parameter .
@xmath58 is the free energy of the superconducting phase ( @xmath59 ) compared to the normal state ( @xmath60 ) which is given by @xcite : @xmath61 , \label{free}\end{aligned}\ ] ] the coefficients @xmath12 and @xmath32 are given by @xmath62 and @xmath63 , where @xmath64 and @xmath65 is the reduced temperature @xmath66 , @xmath67 is the gl parameter , @xmath68 is the pair electric charge .
@xmath69 ( @xmath70 ) is the size of the superconducting domain along the a ( b ) axis .
the josephson couplings are @xmath71 here @xmath72 ( @xmath73 ) is the distance between two next neighboring superconducting domains along the @xmath12 ( @xmath32 ) axis , while @xmath74 is the interplane distance(fig.1 ) . @xmath75 and @xmath76 are the effective pair masses in , respectively , the superconducting domain and along the transverse @xmath33 axis . hereafter , we consider @xmath77 since they correspond to the inplane effective cooper pair masses whereas @xmath76 will be expressed as a function of the anisotropy parameter @xmath78 as : @xmath79 .
( @xmath70 ) in the @xmath12 ( @xmath32 ) direction .
the thickness of the non - superconducting domain is denoted by @xmath72 ( @xmath73 ) along the @xmath12 ( @xmath32 ) axis .
the interplane distance @xmath74 has been exaggerated for clarity .
the intraplane ( @xmath30 and @xmath31 ) and the interplane ( @xmath34 ) josephson couplings are also represented .
] the langevin forces @xmath57 are correlated through the gaussian white - noise law @xcite : @xmath80 with @xmath81 and @xmath82 . assuming a hartree approximation for the quartic term in eq.[free ] leads to a linear problem where the @xmath83 term in the derivative of the free energy is replaced by @xmath84 @xcite .
given the eqs .
[ partial ] and [ free ] , the tdgl equation can be written as : @xmath85 it is worth stressing that the tdgl theory we used was first proposed by puica and lang @xcite to study the effect of the superconducting fluctuations on the inplane conductivity in a self consistent hartree approximation and by mishonov _
et al._@xcite within a gaussian approximation .
moreover , the non - ohmic effect of the high electric field on the transverse magnetoconductivity has been studied in the frame of tdgl by puica and lang @xcite .
the authors have also interpreted , within the same approach , the excess hall conductivity in the layered superconductors for an arbitrarily inplane electric field @xcite .
the tdgl has proven to be a powerful tool to investigate the role of superconducting fluctuations in the critical region to go beyond the usual aslamazov - larkin and lawrence - doniach approaches .
+ the form of the free energy given by eq.[free ] is based on the model proposed by puica and lang @xcite to study the role of critical fluctuations on the conductivity of high-@xmath0 superconductors .
the authors considered a layered structure with only interplane josephson coupling @xmath34 connecting homogeneous superconducting layers .
following the approach of puica and lang @xcite , the critical temperature t@xmath11 of the superconducting state whose free energy is given by eq.[free ] can be derived by solving the equation @xmath86 to solve this equation , we use the green function method @xcite and define the fourier transforms of @xmath48 and @xmath87 by : @xmath88 @xmath89\psi_q(\vec{k},t ) \label{fzeta}\end{aligned}\ ] ] with @xmath90 where @xmath91 and @xmath92 .
equation [ fzeta ] is obtained by taking the fourier transform of eq.[zeta ] .
we define the green function @xmath93 of eq.[fzeta ] by : @xmath94r_q(\vec{k},t , k^{\prime}_x , t^{\prime})= \delta(k_x - k^{\prime}_x)\delta(t - t^{\prime } ) \label{r}\end{aligned}\ ] ] where @xmath95 the solution of eq.[fzeta ] is given by : @xmath96 @xmath97 can be derived by taking the fourier transform of the green function with respect to time @xcite
@xmath98 eq.[r ] can be written , then , as : @xmath99r_q(\vec{k},t , k^{\prime}_x , t^{\prime})= \delta(k_x - k^{\prime}_x ) \label{fr}\end{aligned}\ ] ] one can then obtain @xmath100 and solve eq.[self ] for @xmath101 to derive the transition temperature t@xmath11 .
straightforward calculations give rise to the following equation to which t@xmath11 obeys : @xmath102 ^ 2\right\rbrace = 0 \label{tc}\end{aligned}\ ] ] where @xmath103 and @xmath104 are cutoff parameters which depend on the energy scales of the system while @xmath105 and the reduced josephson couplings @xmath106 ( @xmath107 ) are given by : @xmath108 from eq.[tc ] , one can derive the dependence of t@xmath11 as a function of the @xmath106 couplings ( fig.1 of ref.@xcite ) .
however , a quantitative comparison with the experimental results requires the determination of the relationship between the josephson couplings and the segregated structure induced by disorder .
this point is discussed in the next section .
+ let us now turn to the effect of the phase segregation on the upper critical field of bechgaard salts .
we focus , for simplicity , on the behavior of the upper critical field along the most conducting axis ( @xmath109 ) to avoid complexity related to the field induced confinement occurring for @xmath110 .
moreover , we will tackle the case of ( tmtfs)@xmath2pf@xmath5 in the vicinity of the critical pressure @xmath7 at which the sdw phase vanishes@xcite . in this region ,
the inhomogeneous structure is well established and the geometry of the domains has been determined experimentally @xcite .
we give in the following , the outlines of the model .
details will be published elswhere .
+ in the slab structure , reported in refs .
@xcite , superconducting domains of thickness @xmath111 are sandwiched by sdw insulating regions along the most conducting axis @xmath12 .
we denote by @xmath112 the size of these sdw domains .
the free energy of the system in the presence of a magnetic field along the @xmath12 direction @xmath113 is given by : @xmath114 , \label{free2}\end{aligned}\ ] ] where @xmath115 is the superconducting order parameter in the @xmath50 slab and @xmath116 is the josephson coupling constant between two neighboring slabs : @xmath117 .
considering the gauge @xmath118 , the tdgl equation becomes : @xmath119\psi_n=\zeta_n(\vec{r},t)\end{aligned}\ ] ] using the method introduced by ullah and dorsey @xcite and by puica and lang@xcite in the case of layered superconductors where superconducting planes are coupled via josephson parameters , one can deduce the upper critical field @xmath120 for a superconductor with a slab structure .
details of calculations will be given in a forthcoming paper .
the results is discussed in the next .
disorder can be generated , as we have seen , by different sources : cooling , anion or donor substitution and irradiation .
regardless of the origin of the introduced disorder , we have to establish the connection between the experimental parameters measuring its amount and the size of the non - superconducting domain on which depend the josephson couplings @xmath30 and @xmath31 given by eq.[j ] .
for simplicity , we assume hereafter that the superconducting clusters are isotropic with a length @xmath111 . this assumption ds not affect the outcomes of the study as we shall show in the following .
the size of the non - superconducting domain is denoted , henceforth , @xmath112 .
pouget _ et al .
_ @xcite have carried out a high resolution structural study on ( tmtsf)@xmath2clo@xmath4 based on x - ray synchrotron radiation diffraction .
the authors determined , for different cooling rates , the average size @xmath121 ( i= @xmath12 , @xmath32 , @xmath33 ) along the crystallographic directions and the fraction @xmath122 of the ordered clo@xmath4 domains where superconductivity develops at low temperature . from these data ,
we have derived the typical size of the non - ordered domains assuming that the ordered domains have a cubic shape with a length @xmath111 separated by non - superconducting slabs of thickness @xmath112 along the three directions .
this 3d structure seems to be in contradiction with the assumption of a layered superconductor considered in our model .
actually , the superconducting phase in organic conductors has a 3d character but regarding the relatively small coherence length @xmath123 along the least conducting axis @xmath33 , this phase is substantially dominated by the inplane parameters .
the coherence lengths in ( tmtsf)@xmath2clo@xmath4 along the crystallographic axes are typically of @xmath124 , @xmath125 and @xmath126 @xcite.@xmath123 amounts to about the interplane distance @xmath127 ( denoted @xmath74 in eq.[j ] ) which supports the assumption of an interlayer josephson coupling whatever the cooling rate . considering the isotropic structure ,
the fraction of the clo@xmath4 ordered domains can be written as : @xmath128 given the experimental data for @xmath122 and @xmath111 , one can deduce the values of @xmath112 in order to have a quantitative comparison between our theoretical results and experiments . in table 1
we have listed , for different cooling rates , the experimental values of the superconducting volume fraction @xmath122 , the lengths of the ordered domains along the crystallographic axes denoted @xmath129 , @xmath130 and @xmath131 and the average length @xmath132 of a superconducting domain within the isotropic assumption .
we also give the typical size @xmath112 of the non - superconducting domain and the renormalized josephson coupling , where @xmath133 ( eqs.[free],[j ] ) .
these values are calculated for an average inplane coherence length @xmath134 . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] [ table5 ]
in figure 2 we plot the superconducting transition temperature t@xmath11 obtained from eq.[tc ] as a function of the cooling rate for ( tmtsf)@xmath2clo@xmath4 salt .
the results are in good agreement with the experimental data of matsunaga _
et al._@xcite and joo _ et al._@xcite .
as shown in figure 2 , t@xmath11 falls with increasing cooling rate but presents a significant slope change at a cooling rate of about 5 k / min above which the effect of the cooling rate on @xmath0 is reduced . for slow cooling ( less than 5 k / min ) , the tiny non - superconducting domains act as local defects which are known , according to the ag law , to have drastic effect on superconductivity .
since epr measurements @xcite revealed the non - magnetic nature of the cooling induced disorder centers in ( tmtsf)@xmath2clo@xmath4 , the sensitivity of t@xmath11 to the cooling rate is a signature of the non - s - wave character of the superconducting order parameter . by quenching ,
the non - superconducting regions get wider and can no more be considered as point defects .
the decrease of t@xmath11 is slowed down and ds no more follow the ag formula .
+ clo@xmath4 .
squares correspond to our results obtained from eq.[tc ] whereas the triangles and filled circles are , respectively , the data of refs.@xcite . lines are guide to eye . ]
the tendency to saturation in this regime reflects the insensitivity of the josephson coupling due to the increasing size of the josephson junction between the superconducting domains .
actually , this saturating regime is expected to end as soon as the size of the superconducting domain becomes smaller than the inplane coherence length leading to the disappearance of the superconducting state .
it is worth noting that , despite the simplifying assumption considered in our model , the obtained results show good agreement with experimental data .
this indicates that actually , in the range of validity of the calculations , the cooling rate acts principally on the size of the non - superconducting domains .
it turns out that the ratio @xmath135 , on which depend the reduced josephson coupling constants @xmath136 in eq.[tc ] , is the key parameter governing the dependence of the superconducting transition on the cooling rate .
+ in figure 3 , we depicted the dependence of t@xmath11 on the dingle temperature corresponding to different cooling rates in @xmath3-(et)@xmath2cu[n(cn)@xmath2]br .
the good agreement between our results and the experimental data @xcite supports the method used to connect the microscopic parameters of our model with the experimental amount of disorder .
it also reflects that the key ingredient is the size @xmath112 of the non - superconducting domains which governs the strength of the josephson coupling .
+ -(et)@xmath2cu[n(cn)@xmath2]br .
squares correspond to the results obtained from eq.[tc ] whereas the filled circles are the data of ref.@xcite .
@xmath137 denotes the value of the dingle temperature at which the experimental data deviate from the ag law . ] in figure 3 we also give the fit of the experimental data with ag formula . at low disorder amount ( small @xmath138 )
, the experimental results are well described by the ag law , indicating that the non - superconducting domains are tiny enough and can be considered as defect points . in this regime ,
our results are also in agreement with the ag formula which means that one recovers the ag law as a limiting case corresponding to small disorder amount . however , for a large disorder amount , there is a clear discrepancy between the ag law and the experimental data since , in this case , the defect point picture ds not hold anymore regarding the increasing size of the non - superconducting domains .
the critical dingle temperature @xmath137 , marked by an arrow in fig.2 , corresponds to a ratio @xmath139 . for @xmath140 ,
the non - superconducting domains are large enough and the ag formula is no more justified . from the value of @xmath137
, one can also deduce the reduced critical value of the mean free path @xmath141 : @xmath142 . + figure 4 ( a ) and ( b ) show the behavior of t@xmath11 with the deuteration rate @xmath29 in @xmath3-[(h@xmath21-et)@xmath9(d@xmath21-et)@xmath143@xmath2cu[n(cn)@xmath2]br for the relaxed and quenched samples respectively .
our results are consistent with the data of yoneyama _ et al .
the calculated t@xmath11 seems to have a decrease rate somewhat greater than the experimental values , in particular for the quenched sample .
this may be due to the simplified assumption concerning the isotropic character of the superconducting domains which we used to derive the length of the non - superconducting regions from the experimental data . with increasing @xmath29 ,
the non - superconducting islands develop resulting in a reduced t@xmath144 .
however , the effect of deuteration induced disorder appears to be rather negligible .
the suppression of t@xmath11 was ascribed by yoneyama _
et al._@xcite to the presence of magnetic impurities .
the non - superconducting domains should be af since the ground state of the d@xmath21-et salt ( @xmath145 ) is af with minor superconducting regions .
for @xmath3-[(h@xmath21-et)@xmath9(d@xmath21-et)@xmath143@xmath2cu[n(cn)@xmath2]br in the relaxed ( a ) and quenched ( b ) samples .
squares correspond to the results obtained from eq.[tc ] and the filled circles are the data of ref.@xcite . ] for @xmath3-[(h@xmath21-et)@xmath9(d@xmath21-et)@xmath143@xmath2cu[n(cn)@xmath2]br in the relaxed ( a ) and quenched ( b ) samples .
squares correspond to the results obtained from eq.[tc ] and the filled circles are the data of ref.@xcite . ]
the suppression of t@xmath11 with irradiation in @xmath3(et)@xmath2cu(ncs)@xmath2 is shown in fig.5 where the residual resistivity @xmath146 measures the irradiation damage .
a fit by the ag formula is indicated by the dashed line .
our results describe correctly the experimental data of analytis _ et al .
_ @xcite and those of sasaki _ et al . _
@xcite , which corroborates the idea that irradiation generate non - superconducting grains in the bulk of the sample and not only at the surface .
moreover , the departure from the ag law can be understood within the present model as a reduction of the josephson tunneling due to increasing size of the non - superconducting regions .
there is no need to assume a multicomponent superconducting order parameter as proposed by analytis _
@xcite , especially since there is no experimental evidence for such scenario .
( et)@xmath2cu(ncs)@xmath2 .
squares correspond to the results obtained from eq.[tc ] and the filled circles are the data of ref.@xcite.,width=302,height=226 ] at the critical value @xmath147 at which the data show a deviation from the ag law , analytis _ et al .
_ @xcite have found that @xmath148 .
this is practically the same value we obtain in the case of @xmath3-(et)@xmath2cu[n(cn)@xmath2]br where @xmath142 .
it seems that there is a critical ratio which limits the range of validity of the ag law in @xmath3-(et)@xmath2x salts .
more experimental data are needed to generalize this finding .
+ the outcome of these results is that our model , despite its simplicity , provides a coherent interpretation of the role of disorder in layered inhomogeneous superconductors .
the technique used to extract the leading microscopic parameter of the model related to the non - superconducting domain length , has proved to be reasonable regarding the good agreement between the calculations and the experimental findings . + figure 6 shows the behavior of the upper critical field along the @xmath12 axis for different rates of disorder assuming the slab structure found in the case of ( tmtsf)@xmath2pf@xmath5 .
the parameter @xmath112 is the size of the non - superconducting slabs . in the case of large disorder amount , corresponding to a large @xmath112 , @xmath1 saturates at a value around the pauli limit ( @xmath149 t ) .
however , by reducing the disorder concentration , which turns out to reduce @xmath112 , @xmath1 is greatly enhanced with a non saturating behavior as observed experimentally @xcite .
we ascribe this feature to the superconducting fluctuations which are enhanced by reducing @xmath112 giving rise to a robust superconducting state .
our results suggest that the superconducting fluctuations are responsible of the non - saturating behavior of @xmath1 in ( tmtsf)@xmath150pf@xmath151 salt .
this finding may shed light on the origin of the discrepancy between the @xmath1 values obtained by thermodynamic and transport measurements in ( tmtsf)@xmath150clo@xmath4@xcite .
the former reveal a saturating upper critical fields whereas the latter show a non saturating behavior .
transport measurements are sensitive to superconducting fluctuations . according to our results ,
the divergent character of the upper critical fields reported by transport measurements is a signature of superconducting fluctuations.regarding the non saturating behavior of the upper critical field , the high field phase has been ascribed to a triplet state whereas the low field phase is assumed to be a singlet phase@xcite .
it is worth to note , that , as done by ullah and dorsey @xcite and puica and lang@xcite , we do not consider in our calculations a particular symmetry of the superconducting state .
the model deals with unconventional superconductors regarding the dependence of the critical temperature on disorder .
however , according to our result , the high field phase could be a singlet state and the divergent character is nothing but the signature of the superconducting fluctuations .
@xmath112 is the thickness of the non superconducting domains and @xmath54 is the coherence length ( @xmath152 ) ] .
we have studied the effect of disorder , regardless of its origin , on the behavior of the superconducting critical temperature @xmath0 in the bechgaard and @xmath3-(et)@xmath2x salts .
we have considered an inhomogeneous superconducting phase where superconducting domains are embedded into a non - superconducting matrix . for small disorder amount
the latter can be considered as defect point and the suppression of @xmath0 is found to be well described by the ag law . however , in the regime of large disorder rate , the non - superconducting regions get wider and the results expected from ag formula are in clear discrepancy with experimental data .
we have then proposed a model to describe the suppression of @xmath0 taking into account the inhomogeneous structure of the disordered material .
we have shown that this texture gives rise to josephson tunneling which depends on the key parameter @xmath135 where @xmath112 is the size of the non - superconducting junction and @xmath54 is the inplane coherence length .
we have also found that , below a critical value @xmath153 of the ratio @xmath154 , where @xmath155 is the inplane mean free path , the ag law holds .
the non - superconducting regions act as impurities in this case .
@xmath153 is found to be of the order of 0.2 in @xmath3 ( et)@xmath2x salts .
however , above @xmath153 , the ag formula is no more reliable and one should consider sizeable non - superconducting domains through which cooper pairs tunnel between superconducting islands .
we have proposed a rather simple method to extract the key microscopic parameters from the experimental data to establish the relationship between our calculations and the experimental data .
a good agreement has been found for various sources of disorder despite the simplifying assumptions .
the model could be improved by including the effect of a random distribution of the superconducting domains as observed recently in the high-@xmath0 superconductors la@xmath2cuo@xmath156 @xcite where a structural ordering of the oxygen interstitials has been reported .
this order , which is highly inhomogeneous , is characterized by a fractal distribution which seems to enhance the superconducting transition temperature @xcite .
such effect should be also relevant in ( tmtsf)@xmath2clo@xmath4 where previous study @xcite has shown the presence of a large distribution of the ordered clo@xmath4 domain size in the relaxed sample .
it has been found that the width of this distribution shrinks as the cooling rate increases @xcite .
we have also studied the role of superconducting fluctuations in layered superconductors with slab structure where transport measurements revealed non saturating upper critical field .
we have ascribed this feature to the enhanced superconducting fluctuations to which transport probes are sensitive .
our results may explain the absence of any divergent behavior in the upper critical fields reported by thermodynamic measurements in organic superconductors .
we warmly thank d. jrome , c. pasquier , j. g. analytis , s. blundell and p. auban - senzier s. yonezawa for helpful and stimulating discussions .
this work was supported by the french - tunisian cmcu project 10 g/1306 .
strack , c. akinci , v. pashchenko , b. wolf , e. uhrig , w. assmus , m. lang , j. schreuer , l. wiehl , j. a. schlueter , j. wosnitza , d. schweitzer , j. mller and j. wykhoff , phys .
b. , * 72 * , 054511 ( 2005 ) .
i. s. beloborodov , k. b. efetov , a. v. lopatin , and v. m. vinokur , rev .
phys . * 79 * , 469 ( 2007 ) , see also _ the physics of superconductors _ vols.i and ii , edited by k.h .
bennemann and j. b. ketterson , springer - verlag ( 2004 ) . | several experimental studies have shown the presence of spatially inhomogeneous phase coexistence of superconducting and non superconducting domains in low dimensional organic superconductors .
the superconducting properties of these systems are found to be strongly dependent on the amount of disorder introduced in the sample regardless of its origin .
the suppression of the superconducting transition temperature @xmath0 shows clear discrepancy with the result expected from the abrikosov - gorkov law giving the behavior of @xmath0 with impurities .
based on the time dependent ginzburg - landau theory , we derive a model to account for the striking feature of @xmath0 in organic superconductors for different types of disorder by considering the segregated texture of the system .
we show that the calculated @xmath0 quantitatively agrees with experiments .
we also focus on the role of superconducting fluctuations on the upper critical fields @xmath1 of layered superconductors showing slab structure where superconducting domains are sandwiched by non - superconducting regions .
we found that @xmath1 may be strongly enhanced by such fluctuations . |
the statistical mechanics paradigm based on boltzmann - gibbs entropy which is great success looks after be incapable to deal with many interesting physical systems@xcite . in 1988 , tsallis advanced a nonextensive generalization of boltzamann - gibbs entropic measure .
there are many applications for tsallis formalism such as the specific heat of the harmonic oscillator @xcite , one - dimensional ising model @xcite , the boltzmann h - theorem @xcite , the ehrenfest theorem @xcite , quantum statistics @xcite , paramagnetic systems @xcite , cercle maps @xcite , henon map @xcite , haldane exclusion statistics @xcite , q - expectation value@xcite , quantum mechanical treatment of constraints@xcite . in this paper
, we will focus on thermodynamics and vibration properties of proteins in the framework of a generalized thermostatistics . before proceeding with generalized gaussian network model
, we described fundamental definitions of tsallis thermostatistics in section ii .
we then study generalized gaussian network model and result in section iii .
results will be discussed in section iv .
the generalization which has been successfully nonextensive statistical mechanics is based on the following expression @xcite @xmath1 in the tsallis statistics , the nonextensive canonical distribution is given by @xmath2^{1/(1-q)},\ ] ] where @xmath3^{1/(1-q)},\end{aligned}\ ] ] is the partition function with @xmath4 and @xmath5 @xmath6 @xmath7 ( with @xmath8 ) .
eq.(2 ) is obtained maximizing the tsallis entropy using lagrange multiply methods@xcite .
the internal energy is described by @xmath9 where @xmath10 in free energy notation , eq.(4 ) appears as @xmath11 where @xmath12 is free energy is described as @xmath13.\end{aligned}\ ] ] the expectation of an observable a is @xmath14
we focus our discussion on the classical tsallis statistics by considering protein folding and harmonic approximation .
consider the following hamiltonian for gaussian network model(gnm ) based on harmonic approximation@xcite @xmath15 where the first term is the kinetic energy of the system and @xmath16 is the strength of the springs . in this model ,
springs are assumed as homogeneous . @xmath17 and
@xmath18 is also defined as the equilibrium position and the displacement with respect to @xmath19 of the @xmath20-th c@xmath21 atoms .
the model is eventually described by the contact matrix @xmath22 with entries : @xmath23 if the distance @xmath24 between two c@xmath21 s , in the native conformation , is below the cutoff @xmath25 , while is @xmath26 otherwise . from eq.(3 ) , partition function for the gaussian network model@xcite which has above hamiltonian is given by @xmath27 ^{1/(1-q)}.\end{aligned}\ ] ] we introduce the variables @xmath28 , @xmath29 and thus @xmath30 if @xmath31 is eigenvalues of the @xmath22 and @xmath32 is eigenvectors of @xmath22 .
we can rewrite partition function as : @xmath33 ^{1/(1-q)}.\end{aligned}\ ] ] to calculate this integral , we definition new variables : @xmath34^{1/2}\ x_{i}$ ] and @xmath35^{1/2}p_{i}$ ] @xcite , where @xmath36 . substituting these variables into @xmath37
gives @xmath38 \right \ } \nonumber \\ & & \times\int \prod_{n=1}^{2n}{\ \mbox d}y_{n}\left ( 1- \sum_{k=1}^{2n}y_{k}^{2}\right ) ^{1/(1-q)}.\end{aligned}\ ] ] by using hyperspherical coordinates with @xmath39 and calculating the integral over the angular variables@xcite , we obtain @xmath40 \right \}\frac{\omega _ { 2n}}{2}\nonumber \\ & & \times \int_{0}^{1}{\mbox d}u\;u^{n-1}\left ( 1-u\right ) ^{1/(1-q)}.\end{aligned}\ ] ] by using definition of the solid angle@xcite @xmath41 and using integral representation of euler beta function@xcite , we obtained that @xmath42 \right \}\frac{\gamma \left ( \frac{1}{1-q}+1\right ) } { \gamma \left ( \frac{1}{1-q}+1+n\right ) } \nonumber \\ & = & \left[\left(\frac{2-q}{1-q}\right)_n\right]^{-1 } \prod_{n=1}^{n}\left [ \frac{2\pi}{(1-q)\lambda_{n}\beta } ( \frac { m}{\gamma})^{1/2}\right],\end{aligned}\ ] ] where @xmath43 is the pochhammer symbol @xcite .
free energy of the system is given by @xmath44 and @xmath45^{-1 } \prod_{n=1}^{n}\left [ \frac{2\pi}{(1-q)\lambda_{n}\beta } ( \frac { m}{\gamma})^{1/2}\right],\end{aligned}\ ] ] and then @xmath45^{-1 } \left [ \frac{2\pi}{(1-q)\beta } ( \frac { m}{\gamma})^{1/2}\right]^{n } ( \det{(l^{-1})})^{1/2}.\end{aligned}\ ] ] from eq.(6 ) , the generalized internal energy can be calculated as @xmath46 ^{n}(\det{(l^{-1})})^{1/2}\left[\left(\frac{2-q}{1-q}\right)_n\right]^{-1 } \right \}^{1-q}.\end{aligned}\ ] ] hence , we obtained thermodynamic properties of the generalized gaussian network model . on the other hand , the most properties of gaussian network model is called temperature factor . the crystallographic temperature factors or b - factors , which is defined as intrinsic fluctuations of the atoms in crystal ,
is directly related to this fluctuations .
the x - ray crystallographic temperature factors are used by the mean square fluctuation of c@xmath47 atoms around their native positions@xcite .
the generalized temperature factor is expressed as @xmath48 with @xmath49 indicating the thermal average . where @xmath50 is defined by @xmath51^\frac{q}{1-q},\end{aligned}\ ] ] or @xmath52^\frac{q}{1-q}.\end{aligned}\ ] ] in the ggnm ( generalized gaussian network model ) approximation , this average is easily carried out , because amounts to a nongaussian integration , and b - factors can be expressed in terms of the diagonal part of the inverse of the matrix @xmath53\end{aligned}\ ] ] . substituting eq .
20 into eq.17 gives @xmath54.\end{aligned}\ ] ] if @xmath0 , it is calculated that same result of the gaussian network model as@xcite @xmath55\end{aligned}\ ] ] for the numerical results , we choice a reasonable cutoff distance including all residues pairs within a first interaction shell is 7.0 .
we display in fig.[eps1 ] temperature factors for t4 lysozyme ( or 3lzm pdb ) as a function of residue index at q=0.99 .
bold and dashed curves depict the experimental and theoretical results , respectively .
the theoretical curve is normalized by taking @xmath56 , therefore as to match the area enclosed by the two curves .
the agreement between the theoretical and experimental curves is remarkable . in fig.[eps2 ] , we also display to behavior cross correlation between i.residues and j.residues of all modes at q=0.99 .
in this study , we obtained temperature factor or beta factor using by tsallis statistical mechanics . in this framework , by the approximate scheme which we have used here , closed and analytical expressions for the generalized gaussian network model have been derived for harmonic oscillator .
in addition to this , the temperature factor is q dependent . since the temperature factor must be positive
, q must be less than unity .
this also shows that q must be less than unity for harmonic oscillator system .
as can be seen in fig.[eps1 ] , at q=0.99 , the beta factors which are obtained from experimental and numerical calculation are in agreement with each other .
the more the q parameter decreases , the more the temperature factor disagrees with experimental data
. therefore , for q values in the lower vicinity of 1 , the experimental data are in agreement with numerical calculations . in this system , as @xmath0 , it is observed that , both of the above expressions reduce to the results obtained in the standard boltzmann - gibbs statistics .
h.a . acknowledges support by the tbitak under the project number 104t150 and from tba under the programme to reward successful young scientists .
e. curando and c. tsallis , j. phys .
a * 24 * , l69 ( 1991 ) ; corrigenda : * 24 * , 3187 ( 1991 ) and * 25 * , 1019 ( 1992 ) .
t. halilolu , i. bahar and b. erman , phys.rev.lett . * 79*,3090 ( 1997 ) .
e. k. lenzi , r. s. mendes , l. r. da silva and l. c. malacarne , phys .
a * 44 * , 289 ( 2001 ) . | the vibrational dynamics of protein folding is analyzed in the framework of tsallis thermostatistics .
the generalized partition functions , internal energies , free energies and temperature factor ( or debye - waller factor ) are calculated .
it has also been observed that the temperature factor is dependent on the non - extensive parameter q which behaves like a scale parameter in the harmonic oscillator model . as @xmath0
, we also show that these approximations agree with the result of gaussian network model .
keywords : tsallis thermostatistics , harmonic oscillator , protein folding , temperature factor |
in a previous article ( see @xcite ) , by introducing the `` similarity '' relation and applying it to the restricted three - body problem , the `` similar '' equations of motion were obtained .
these equations were connected with the classical equations of motion by some coordinate transformation relations ( see equations ( 17 ) in @xcite ) . in this paper
were also defined similar parameters and physical quantities , and similar initial conditions and some trajectories of the test particles into the physical @xmath0 and respectively @xmath1 planes were ploted . denoting @xmath2 and @xmath3 the components of the binary system ( whose masses are @xmath4 and @xmath5 ) , the equations of motion of the test particle ( in the frame of the restricted three - body problem ) in the coordinate system @xmath6 are ( see equations ( 11)-(13 ) in @xcite ) : @xmath7 @xmath8 @xmath9 where @xmath10 in the similar coordinate system @xmath11 the equations of motion of the test particle are ( see equations ( 14)-(16 ) in @xcite ) : @xmath12 @xmath13 @xmath14 where @xmath15 it can be easily verify that the equations of coordinate transformation are : @xmath16 one can observes that equations ( 1)-(3 ) and ( 5)-(7 ) have singularities in @xmath17 , @xmath18 , @xmath19 and @xmath20 .
these situations correspond to collision of the test particle whith @xmath2 or @xmath3 in a straight line .
the collision is due to the nature of the newtonian gravitational force ( @xmath21 ) .
if the test particle approaches to one of the primaries very closely ( @xmath22 ) , then such an event produces large gravitational force ( @xmath23 ) and sharp bends of the orbit .
the removing of these singularities can be done by regularization .
( remark : the purpose of regularization is to obtain regular equations of motion , no regular solutions . )
euler seems to be the first ( in 1765 ) to propose regularizing transformations when studying the motion of three bodies on a straight line ( see @xcite ) .
the regularization method has become popular in recent years ( see @xcite ) for long term studies of the motion of celestial bodies .
these problems have a special merit in astronomy , because with their help we can studied more efficient the equations of motions with singularities . at the collision
the equations of motion possess singularities .
the problem of singularities plays an important role under computational , physical and conceptual aspects ( see @xcite ) .
the singularities occurring at collisions can be eliminated by the proper choice of the independent variable .
the basic idea of regularization procedure is to compensate for the infinite increase of the velocity at collision .
for this reason , a new independent variable , fictitious time , is adopted .
the corresponding equations of motion are regularized by two transformations : the time transformation and the coordinate transformation .
the most important part of the regularization is the time transformation , when a new fictitious time is used , in order to slow the motion near the singularities .
the regularization can be local or global .
if a local regularization is done , then the time and the coordinates transformations eliminate only one of the two singularities .
an example for the local regularization is the birkhoff s transformation ( see @xcite ) .
the global regularization eliminates both singularities at once ( see @xcite ) . because our singularities are given in terms of @xmath24 , @xmath25 , @xmath26 , @xmath27 , in this paper a global regularization
will be done . in order to do this ,
we need to replace the cartesian equations ( 1)-(3 ) and ( 5)-(7 ) with the corresponding canonical equations of motion .
the canonical coordinates are formed by generalized coordinates @xmath28 and generalized momenta @xmath29 .
the hamiltonian , defined by the equation : @xmath30 will becomes ( see @xcite p. 266
, for the generalized momenta , when the coordinates system rotates ) : @xmath31 here @xmath32\;,\ ] ] with @xmath33 here the generalized coordinates and the generalized momenta were : @xmath34 then , the canonical equations @xmath35 have , in the @xmath36 coordinate system , the explicit forme : @xmath37 it is easy to verify that using the relations ( 14 ) , the explicit canonical equations become the cartesian equations ( 1)-(3 ) . in order to write the canonical equations in the similar
coordinate system @xmath38 , we have in view the theoretical considerations from the article @xcite .
the index _ s _ refers to similar quantities .
then , the similar hamiltonian will be : ( see @xcite p. 266 ) : @xmath39 where @xmath40\;,\ ] ] with @xmath41 here the generalized coordinates and the generalized momenta were : @xmath42 then , the canonical equations @xmath43 have , in the @xmath38 coordinate system , the explicit forme : @xmath44 it is easy to verify that using the relations ( 27 ) , ( 28 ) , ( 29 ) , the explicit canonical equations ( 30 ) , ( 31 ) , ( 32 ) , become the cartesian equations ( 5)-(7 ) . _
remark _ : from equations ( 13 ) and ( 24 ) it is easy to observe that @xmath45 and @xmath46 ( see also figure 1 in @xcite ) .
the equations of motion ( 19)-(21 ) and ( 30)-(32 ) have singularities in @xmath47 and @xmath48 , respectively in @xmath49 and @xmath50 . we shall remove these singularities by regularization .
several regularizing methods are known ( see stiefel et al . 1971 ) . in this paper
we shall use the levi - civita s method , applied when the bodies are moving on a plane .
the two steps performed in the process of regularization of the restricted problem are the introduction of new coordinates and the transformation of time .
the combination of the coordinate ( dependent variable ) transformation and the time ( independent variable ) transformation have an analytical importance and increase the numerical accuracy . for simplicity we shall consider that the third body moves into the orbital plane .
for the regularization of the equations of motion in the @xmath51 coordinate system , we shall introduce new variables @xmath52 and @xmath53 , conected with the coordinates @xmath54 and @xmath55 by the relations of levi - civita ( see @xcite ) : @xmath56 let introduce the generating function @xmath57 ( see @xcite , p.196 ) : @xmath58 a twice continuously differentiable function . here
@xmath59 and @xmath60 are harmonic conjugated functions , with the property @xmath61 the generating equations are @xmath62 with @xmath63 as new generalized momenta , or explicitly @xmath64 where @xmath65 let introduce the following notation : @xmath66 @xmath67 where @xmath68 represents the transpose of matrix @xmath69
. the new hamiltonian with the generalized coordinates @xmath52 and @xmath53 and generalized momenta @xmath70 and @xmath71 is : @xmath72 + \frac{q}{1+q}f- \nonumber \\ & - & \frac{1}{1+q}\cdot \frac{1}{\overline{r}_1 } - \frac{q}{1+q } \cdot \frac{1}{\overline{r}_2}-\frac{q^2}{2(1+q)^2}\end{aligned}\ ] ] where @xmath73 , @xmath74 and @xmath75 and the explicit canonical equations of motion in new variables become : @xmath76 \nonumber\\ \frac{dq_2}{dt}&=&\frac{1}{2d } \left [ 2p_2-\frac{\partial } { \partial q_1 } ( f^2+g^2 ) \right ] \nonumber\\ \frac{dp_1}{dt}&=&- \frac{p_1}{2d } \cdot \frac{\partial } { \partial q_1\partial q_2 } ( f^2+g^2 ) + \frac{p_2}{2d } \cdot \frac{\partial } { \partial q_1 \partial q_1 } ( f^2+g^2 ) - \frac{q}{1+q } \frac{\partial f}{\partial q_1}+ \nonumber \\ & + & \frac{1}{1+q}\cdot \frac{\partial } { \partial q_1 } \left ( \frac{1}{\overline{r}_1 } \right ) + \frac{q}{1+q } \cdot \frac{\partial } { \partial q_1 } \left ( \frac{1}{\overline{r}_2 } \right ) \nonumber \\ \frac{dp_2}{dt}&=&- \frac{p_1}{2d } \cdot \frac{\partial } { \partial q_2\partial q_2 } ( f^2+g^2 ) + \frac{p_2}{2d } \cdot \frac{\partial } { \partial q_2 \partial q_1 } ( f^2+g^2 ) - \frac{q}{1+q } \frac{\partial f}{\partial q_2}+ \\ & + & \frac{1}{1+q}\cdot \frac{\partial } { \partial q_2 } \left ( \frac{1}{\overline{r}_1 } \right ) + \frac{q}{1+q } \cdot \frac{\partial } { \partial q_2 } \left ( \frac{1}{\overline{r}_2 } \right)\nonumber\end{aligned}\ ] ] using levi - civita s transformation @xmath77 ( see relations ( 33 ) ) , the equations ( 39 ) becomes : @xmath78 where @xmath79 , @xmath80 , + with the new hamiltonian @xmath81 to resolve the hamiltonian equations ( [ eq1 - 14 ] ) , we introduce the fictitious time @xmath105 , ( see @xcite ) , and making the time transformation @xmath106 , the new regular equations of motion are @xmath107 the explicit equations of motion may be written @xmath108 _ remark _ : it is easy to see that now , the equations of motion have no singularities . for the application of the above problem in a binary system , we can obtain the solution in the form @xmath109
for the coordinate transformation in the @xmath82 coordinate system , we introduce the generating function @xmath83 in the plane @xmath84 , in the following form @xmath85 where @xmath86 and @xmath87 are harmonic conjugated functions .
the generating equations are @xmath88 or explicitly @xmath89 where @xmath90 let introduce the following notation , ( @xcite , p. 373 ) @xmath91 @xmath92 the new hamiltonian for the case 2 may be written @xmath93 + \frac{q'}{1+q'}f_s- \nonumber \\ & - & \frac{1}{1+q'}\cdot \frac{1}{\overline{r}_{s1 } } - \frac{q'}{1+q ' } \cdot \frac{1}{\overline{r}_{s2}}-\frac{q'^2}{2(1+q')^2}\end{aligned}\ ] ] where @xmath94 , @xmath95 and @xmath96 and the hamiltonian equations in new variables become @xmath97 \nonumber\\ \frac{dq_{s2}}{dt}&=&\frac{1}{2d_s } \left [ 2p_{s2}+\frac{\partial } { \partial q_{s1 } } ( f_s^2+g_s^2 ) \right ] \nonumber\\ \frac{dp_{s1}}{dt}&= & \frac{p_{s1}}{2d_s } \cdot \frac{\partial } { \partial q_{s1}\partial q_{s2 } } ( f_s^2+g_s^2 ) - \frac{p_{s2}}{2d_s } \cdot \frac{\partial } { \partial q_{s1 } \partial q_{s1 } } ( f_s^2+g_s^2 ) - \frac{q'}{1+q ' } \frac{\partial f_s}{\partial q_{s1}}+ \nonumber \\ & + & \frac{1}{1+q'}\cdot \frac{\partial } { \partial q_{s1 } } \left ( \frac{1}{\overline{r}_{s1 } } \right ) + \frac{q'}{1+q ' } \cdot \frac{\partial } { \partial q_{s1 } } \left ( \frac{1}{\overline{r}_{s2 } } \right ) \nonumber \\ \frac{dp_{s2}}{dt}&= & \frac{p_{s1}}{2d_s } \cdot \frac{\partial } { \partial q_{s2}\partial q_{s2 } } ( f_s^2+g_s^2 ) - \frac{p_{s2}}{2d_s } \cdot \frac{\partial } { \partial q_{s2 } \partial q_{s1 } } ( f_s^2+g_s^2 ) - \frac{q'}{1+q ' } \frac{\partial f_s}{\partial q_{s2}}+ \nonumber \\ & + & \frac{1}{1+q'}\cdot \frac{\partial } { \partial q_{s2 } } \left ( \frac{1}{\overline{r}_{s1 } } \right ) + \frac{q'}{1+q ' } \cdot \frac{\partial } { \partial q_{s2 } } \left ( \frac{1}{\overline{r}_{s2 } } \right).\end{aligned}\ ] ] because the singularity of the problem is given by the terms @xmath98 and @xmath99 , we will made a global regularization using the levi - civita s transformation @xmath100 the similar hamiltonian equations are given by @xmath101 where @xmath102 , @xmath103 , + with the new hamiltonian @xmath104 introducing the fictitious time @xmath105 and making the time transformation @xmath110 , the new regular equations of motion are obtained in the form @xmath111 the explicit equations of motion are given by @xmath112 _ remark _ : it is easy to see that the similar equations of motion have no singularities .
for the application of the above problem in a binary system , we can obtain the solution in the form @xmath113
the transformation of the independent variable is necessary to achieve regularization .
it is a slow - down treatment of the physical problem , a new time scale in which the motion slows down @xcite .
for the numerical integration ( earth - moon binary system ) , considering that the third body moves into the orbital plane ( see @xcite ) , we used the initial values : @xmath114 , \;\;q=0.0123 \;.\ ] ] for the numerical integration ( earth - moon binary system ) in the `` similar '' coordinate system we use the initial values ( see eqs . ( 25 ) ) : @xmath115,\;\ ; q'=81.30\;.\ ] ] for the numerical integration ( earth - moon binary system ) in the regularized coordinate system ( equations ( 52 ) ) , we use the initial values ( see also eq . ( 33 ) ) : @xmath116,\;\ ; q=0.0123 \;,\ ] ] and in the similar regularized coordinate system ( equations ( 54 ) ) : @xmath117,\;\ ; q'=81.30\;.\ ] ] in figure 1 we can compare the trajectories of the test particle in the coordinate systems with origin in @xmath2 ( figures _ a _ , _ c _ , _ e _ ) , and @xmath3 ( figures _ b _ , _ d _ , _ f _ ) .
the point @xmath70 correspond to the initial conditions .
we consider the trajectories given in @xcite in figure 6 ( in the coordinate systems @xmath0 and @xmath118 ) and we represented them in the coordinate systems @xmath119 and @xmath120 ( see figure 1 _ a _ and _ b _ ) . in this purpose
we obtained the initial conditions as follows : @xmath121 and from eqs .
( 16)-(17 ) : @xmath122 and @xmath123 and from eqs .
( 27)-(28 ) : @xmath124 in order to obtain the initial conditions , when we make the coordinate transformation , we solve the systems : @xmath125 , @xmath126 + ( see eqs .
( 33 ) and ( 36 ) ) for the trajectory in @xmath127 coordinate system ( figure 1c ) and @xmath128 , @xmath129 + ( see eqs .
( 44 ) for the trajectory in @xmath130 coordinate system ( figure 1d ) .
obviously , the initial conditions remain the same if we change the real time _
t _ to the fictitious time @xmath105 , but the motion is slowed . in figure 1e and figure 1f
we represented the motion in real time _
t _ with thin line and the slowed motion with thick line ( corresponding to the same period of time ) .
let us analyze the figures 1a and 1c .
for this purpose we consider a point a(@xmath131 ) on the graphic show in figure 1a , and b(@xmath132 ) its corresponding point in figure 1c .
we have ( see figure 2a and 2b ) : @xmath133 and it results : @xmath134 .
we used the counterclockwise directions for measuring the angles .
the levi - civita geometrical transformation originate in the conformal transformation ( see @xcite , p.164 ) : @xmath135 where @xmath136 is the physical plane and @xmath127 is the parametric plane . from this relation
we have : @xmath137 , @xmath138 , and @xmath139 .
it means that the geometrical transformation squares the distances from the origin and doubled the polar angles .
if , having the trajectory in the physical plane , we want to draw the trajectory into the parametric plane , we have to choose a point @xmath140 on the trajectory in @xmath136 plane , measure the angle @xmath141 and the distance @xmath142 , and then draw a half - line @xmath143 in the @xmath144 plane , so as @xmath145 . on this half - line , we have to measure the distance @xmath146 , and obtain the point @xmath147 . than we have to repeat the procedure for @xmath148 .
of course the computer will do this better and faster than we can do it , but the above considerations help us to understand what it happened .
the vertex of the polar angles have to be centered into the more massive star , so the angles @xmath149 and @xmath150 and respectively @xmath151 and @xmath152 are similar polar angles .
so , if we intend to study the regularization of the circular restricted three - body problem using similar coordinate systems , we have to add to similar parameters postulated in section 3 in ( roman , 2011 ) , _ the similar polar angles _ , measured between the abscissa and the half - line passing through the center of the most massive star and the test particle .
+ in order to see what is the effect of geometrical transformation , let us analyze the graphics from figure 3 . in figure 3a there
are represented some circles ; their equations are : @xmath153 , where @xmath154 . in figure 3b
there are represented the circles having the equations @xmath155 , where @xmath156 , like geometrical transformation of levi - civita s regularization postulated .
one can see that the circles in figure 3b go away from the center and draw near the circle having radius @xmath157 .
if in the center of the circles there is a problem ( a singularity ) , it can be easily examined . in figure 3c and figure
3d there are represented some half - lines , having equations : @xmath158 , respectively @xmath159 , where @xmath160 , and @xmath161 , like geometrical transformation of levi - civita s regularization postulated .
one can see that the half - lines in figure 3d go away from the abscissa s axis .
if there is a problem ( a singularity ) on the abscissa s axis , it can be easily examined .
there are only two points invariant with respect to the geometrical transformation of levi - civita s regularization : @xmath162 , and @xmath163 , respectively in the similar coordinate systems @xmath164 , and @xmath165 .
then , the geometrical transformation go away the trajectory from the points where there are singularities . in what concern the time transformation , as one can see in figure 1e and 1f ,
the role of this transformation is to slow - down the motion of the test particle . with thin line
is represented the trajectory of the test particle when the time integration is @xmath166 , and with thick line is represented the trajectory of the test particle when the time integration is @xmath167 .
as one can see , after @xmath167 we are still far away from the point where it is possible to have a singularity , if the coordinate system has the origin in @xmath2 , but not so far away if the origin of the coordinate system is in @xmath3 .
this paper continue the study of the relation of similarity , postulated in @xcite , by applying it to the levi - civita s regularization of the motion s equations of the test particle , in the circular , restricted three - body problem .
many papers in the last decade have studied the restricted three - body system in a phase space . during these studies ,
difficulties have arisen when the system approaches a close encounter .
using the regularization method in the similar coordinates system , we give explicitly equations of motion for the test particle .
we study numerically the regular equations of motion , we written in canonical form , and obtained that the integrator using regularized equations of motion are more efficient .
the similar hamiltonian ( see eq . ( 22 ) ) give us the similar canonical equations ( 27)-(32 ) , which have some different signes than the canonical equation ( 16)-(21 ) .
the coordinate transformation used in the levi - civita s regularization create a new form of the similar hamiltonian s equations ( eqs .
finally , the time transformation used in the levi - civita s regularization gives us the regularized equations of motion ( 51 ) in the coordinate system with origin in @xmath2 , and ( 54 ) in the similar coordinate system .
birkhoff , g.d .
cir . mat .
palermo , 39,1 ( 1915 ) boccaletti , & d. , pucacco , g. : theory of orbits , vol.1 , springer - verlag berlin heidelberg new york ( 1996 ) castilho , c. , & vidal , c. : qual .
theory dyn .
syst . , 1 , 1 ( 1999 )
celletti , a. , stefanelli , l. , lega , e. , & froeschl , c. : celest .
, 109 , 265 ( 2011 ) csillik , i. : regularization methods in celestial mechanics , cluj : house of the book of science ( 2003 ) rdi , b. : celest .
astron . , 90 , 35 ( 2004 ) jimnez - perez , h. , & lacomba , e. : j. phys . a , 44 , 265 ( 2011 ) kopal , z. : dynamics of close binary systems , reidel , dordrecht ( 1978 ) levi - civita , t. : acta math . 30 , 305 - 327 ( 1906 ) mikkola , s. , & aarseth , s.j .
astron . , 64 , 197 ( 1996 )
mioc , v. , & csillik , i. : roaj , 12 , 167 ( 2002 ) roman , r. : , doi : 10.1007/s10509 - 011 - 0747 - 1 ( 2011 ) stiefel , l. , & scheifele , g. : linear and regular celestial mechanics , berlin : springer ( 1971 ) szebehely , v. : theory of orbits . academic press , new york ( 1967 ) szebehely , v. : regularization in celestial mechanics , in lecture notes in mathematics , vol .
461 , p.257 - 263 ( 1975 ) waldvogel , j. : celest .
astron . , 6 , 221 ( 1972 )
waldvogel , j. : celest .
, 28 , 69 ( 1982 ) waldvogel , j. : celest .
astron . , 95 , 201 ( 2006 ) | the regularization of a new problem , namely the three - body problem , using similar coordinate system is proposed . for this purpose
we use the relation of similarity , which has been introduced as an equivalence relation in a previous paper ( see @xcite ) .
first we write the hamiltonian function , the equations of motion in canonical form , and then using a generating function , we obtain the transformed equations of motion . after the coordinates transformations , we introduce the fictitious time , to regularize the equations of motion .
explicit formulas are given for the regularization in the coordinate systems centered in the more massive and the less massive star of the binary system .
the similar polar angle s definition is introduced , in order to analyze the regularization s geometrical transformation .
the effect of levi - civita s transformation is described in a geometrical manner . using the resulted regularized equations , we analyze and compare these canonical equations numerically , for the earth - moon binary system . |
weak gibbs measures are conformal probability measures obtained as eigenvectors of ruelle - perron - frobenius operators associated with continuous potentials on topological dynamical systems . when the system @xmath3 has nice enough geometric properties , for instance in the case of a conformal repeller , these measures provide natural , and now standard examples of measures obeying the multifractal formalism : their hausdorff spectrum and @xmath4-spectrum form a legendre pair .
specifically , for such a measure @xmath5 on @xmath3 , the ( lower ) @xmath4-spectrum @xmath6 is defined by @xmath7 where the supremum is taken over all families of disjoint closed balls @xmath8 of radius @xmath9 with centers in @xmath10 ; the hausdorff spectrum of @xmath5 is defined by @xmath11 where @xmath12 stands for the hausdorff dimension , @xmath13 with @xmath14 and we have the duality relation @xmath15 a negative dimension meaning that the set is empty .
in fact , due to the super and submultiplicativity properties associated with @xmath5 , the same equality holds if we replace the limit by a @xmath16 or a @xmath17 in the definition of the local dimension .
the rigorous study of these measures started with the gibbs measures case , which corresponds to hlder continuous potentials , or continuous potentials possessing the so - called bounded distorsions property , and in particular on the so - called `` cookie - cutter '' cantor sets associated with a @xmath18 expanding map @xmath19 on the line @xcite ( see @xcite for an extended discussion of dimension theory and multifractal analysis for hyperbolic conformal dynamical systems ) .
this followed seminal works by physicists of turbulence and statistical mechanics pointing the accuracy of multifractals to statistically and geometrically describe the local behavior of functions and measures @xcite . in the case of gibbs measures ,
the @xmath4-spectrum of the gibbs measure is differentiable , and analytic if the potential @xmath20 is hlder continuous ; it is the unique solution @xmath21 of the equation @xmath22 , where @xmath23 stands for the topological pressure .
the general case of continuous potentials was solved later in @xcite , with the same formula for the @xmath4-spectrum .
these progress then led to the multifractal analysis of bernoulli convolutions associated with pisot numbers @xcite .
thermodynamic formalism and large deviations are central tool in these studies . in the context of random dynamical systems ,
the multifractal analysis of random gibbs measures ( to be defined below ) associated with random hlder continuous potentials on attractors of random @xmath18 expanding ( or expanding in the mean ) random conformal dynamics encoded by random subshifs of finite type has been studied in @xcite , @xcite and @xcite .
these works , as well as the dimension theory of attractors of random dynamics @xcite , are based on the thermodynamic formalism for random transforms @xcite .
the multifractal analysis of random weak gibbs measures is also implicitly considered in @xcite ( which deals with the multifractal analysis of birkhoff averages ) , but the fibers are deterministic , and the techniques developed there seems difficult to adapt in a simple way in the case of random subshifts . in this paper
we consider , on a base probability space @xmath24 , random weak gibbs measures on some class of attractors included in @xmath25 $ ] and associated with @xmath0 random dynamics semi - conjugate ( up to countably many points ) , or conjugate , to a random subshift of finite type .
we provide a study of the multifractal nature of these measures , including the validity of the multifractal formalism , the calculation of hausdorff and packing dimensions of the so - called level sets of divergent points , and a @xmath1-@xmath2 law for the hausdorff and packing measures of the level sets of the local dimension .
compared to the above mentioned works , apart the source of new difficulties coming from the relaxation of the regularity properties of the potentials , our assumptions provide a more general process of construction of the random cantor set in terms of the distribution of the random family of intervals used to refine the construction at a given step : it can contain contiguous intervals ( i.e. without gap in between , and even no gap ) with positive probability ; thus , for instance , it covers the natural families of cantor sets one can obtain by picking at random a fiber in a bedford - mcmullen carpet .
extensions of our results to the higher dimensional case will be discussed in remark [ higherdim ] .
we focus on the one dimensional case because our model will be used in a companion paper to study the multifractal nature of discrete measures obtained as `` inverse '' of the random weak gibbs measures considered here .
section 2 developes background about random dynamical systems and thermodynamic formalism , and presents our main results , namely theorems [ multifractal initial ] and [ multifractal initial2 ] , as well as concrete examples of random attractors .
section 3 provides the basic properties that will used in the proof of this theorem in section 4 .
we first need to expose basic facts from random dynamical systems and thermodynamic formalism .
* random subshift . *
let @xmath26 be a complete probability space and @xmath27 a @xmath28-preserving ergodic map .
let @xmath29 be a @xmath30 valued random variable ( r.v . )
such that @xmath31 and @xmath32 let @xmath33 be a random transition matrix such that @xmath34 is a @xmath35-matrix with entries 0 or 1 .
we suppose that the map @xmath36 is measurable for all @xmath37 and each @xmath34 has at least one non - zero entry in each row and each column .
let @xmath38 and @xmath39 be the left shift @xmath40 for any @xmath41 .
define @xmath42 and the map @xmath43 as @xmath44 .
define the map @xmath45 as @xmath46 .
the corresponding family @xmath47 is called a random subshift .
we assume that this random subshift is topologically mixing , i.e. there exists a @xmath30-valued r.v .
@xmath48 on @xmath26 such that for @xmath28-almost every ( a.e . )
@xmath49 , @xmath50 is positive . for each @xmath51 ,
define @xmath52 as the set of words @xmath53 of length @xmath54 , also denoted @xmath55 , such that @xmath56 for all @xmath57 and @xmath58 for all @xmath59 . for @xmath60 , we define the cylinder @xmath61_{\omega}$ ] as @xmath61_{\omega}:=\{{\underline{w}}\in\sigma_\omega : w_i = v_i\text { for } i=0,\dots , n-1\}.$ ] for any @xmath62 , @xmath63 and @xmath64 , there is at least one word @xmath65 such that @xmath66 . for each such connection
, we fix one such @xmath67 and denote the word @xmath68 by @xmath69
. * relativized entropy . * using the same notations as in @xcite ,
let @xmath70 and @xmath71 any @xmath72 on @xmath73 disintegrates in @xmath74 where the measures @xmath75 , are regular conditional probabilities with respect to the @xmath27-algebra @xmath76 , where @xmath77 is the canonical projection from @xmath78 to @xmath79 .
this implies that for @xmath28-almost every @xmath49 , for any measurable set @xmath80 , @xmath81 , where @xmath82 .
let @xmath83 be a finite or countable partition of @xmath73 into measurable sets .
then for all @xmath84 , @xmath85 is a partition of @xmath86 .
given @xmath87 , the conditional entropy of @xmath88 given @xmath76 is defined by @xmath89 where @xmath90 denotes the usual entropy of a partition @xmath91 .
now , given a finite or countable partition @xmath92 of @xmath78 , define the fiber entropy of @xmath93 , also called the relative entropy of @xmath93 with respect to @xmath92 , as @xmath94 ( here @xmath95 denotes the join of partitions ) .
then define @xmath96 where the supremum is taken over all finite or countable measurable partitions @xmath97 of @xmath73 with finite conditional entropy , that is @xmath98 . in our setting , we have @xmath99 .
the number @xmath100 , also denoted @xmath101 in the literature , is the relativized entropy of @xmath93 given @xmath102 .
it is also called the fiber entropy of the bundle random dynamics @xmath93 .
* topological pressure and random weak gibbs measures .
* we say that a measurable function @xmath103 on @xmath73 is in @xmath104 if 1 .
@xmath105 + where @xmath106 , 2 . for @xmath28-a .
e. @xmath49 , @xmath107 as @xmath108 , where @xmath109 now , if @xmath110 , due to kingsman s subadditive ergodic theorem , @xmath111_\omega}\exp\left(s_n\phi(\omega,\underline v)\right ) \ ] ] exists for @xmath28-a.e .
@xmath49 and does not depend on @xmath49 , where @xmath112 .
it is called _ topological pressure of @xmath103 _ and is denoted @xmath113 .
also , with @xmath103 is associated the ruelle - perron - frobenius operator @xmath114 defined as @xmath115 [ eigen]@xcite removing from @xmath79 a set of @xmath28-probability 0 if necessary , for all @xmath116 there exists @xmath117 and a probability measure @xmath118 on @xmath86 such that @xmath119 .
we call the family @xmath120 a random weak gibbs measure on @xmath121 associated with @xmath103 . for any @xmath84 ,
let @xmath122,u_{\omega}^2=[a_{\omega,2},b_{\omega,2}],\cdots u_{\omega}^{s}=[a_{\omega , s},b_{\omega , s}]\cdots$ ] be closed non trivial intervals with disjoint interiors and @xmath123 .
we assume that for each @xmath124 , @xmath125 is measurable , as well as @xmath126 and @xmath127 .
let @xmath128 and consider a measurable mapping @xmath129 from @xmath130 to the space of @xmath0 diffeomorphisms of @xmath25 $ ] endowed with its borel @xmath27-field .
we consider the measurable @xmath0 diffeomorphism @xmath131 $ ] by @xmath132 .
we denote the inverse of @xmath133 by @xmath134 .
we also define @xmath135),\ \forall v = v_0v_1\cdots v_{n-1}\in\sigma_{\omega , n},\\ x_{\omega}&=&\bigcap_{n\geq 1}\bigcup_{v\in\sigma_{\omega , n}}u_{\omega}^v,\\ x_{\omega}&=&\{(\omega , x):\omega\in\omega , x\in x_{\omega}\},\end{aligned}\ ] ] and for all @xmath84 , @xmath136 and @xmath137 , @xmath138 we say that a measurable function @xmath139 defined on @xmath140 is in @xmath141))$ ] if 1 .
@xmath142 , where @xmath143 , 2 . for @xmath28-a.e .
@xmath84 , @xmath144 as @xmath145 , where @xmath146 we will make the following assumption : @xmath147))$ ] and @xmath148 satisfies the contraction property in the mean @xmath149 under this assumption , there is @xmath28-almost surely a natural projection @xmath150 defined as @xmath151 this mapping may not be injective , but any @xmath152 has at most two preimages in @xmath86 .
we can define @xmath153 , then @xmath154 . by a standard way
, we can easily get that for @xmath28-almost every @xmath116 , the bowen - ruelle formula holds , i.e. @xmath155 where @xmath156 is the unique root of the equation @xmath157 .
our results statement require some additional definitions related to multifractal formalism .
let @xmath5 be a compactly supported positive and finite borel measure on @xmath158 .
the ( lower ) @xmath4-spectrum @xmath6 and the upper-@xmath4 spectrum @xmath159 are respectively defined by @xmath160 where the supremum is taken over all families of disjoint closed balls @xmath8 of radius @xmath9 with centers in @xmath10 . by construction , the function @xmath161 is non decreasing and concave over its domain , which equals @xmath162 of @xmath163 ( see @xcite ) .
the lower and upper large deviations spectra @xmath164 and @xmath165 are given by @xmath166 @xmath167 where the supremum is taken over all families of disjoint closed balls @xmath168 of radius @xmath9 with centers @xmath169 in @xmath170 . for all @xmath171 , define @xmath172 then , for @xmath173 , define @xmath174 it is clear that since @xmath5 is bounded , @xmath175 if @xmath176 . finally , define @xmath177 and @xmath178 equivalent definitions are ( see @xcite ) : @xmath179and @xmath180 ( legendre transform ) for any function @xmath181 with non - empty domain , its legendre transform @xmath182 is defined on @xmath183 by @xmath184 one always has ( see @xcite ) @xmath185 and ( see @xcite ) @xmath186 ( multifractal formalism ) we say that @xmath5 obeys the multifractal formalism at @xmath187 if @xmath188 , and that the multifractal formalism holds ( globally ) for @xmath5 if it holds at any @xmath187 ( here a negative dimension means that the set is empty ) .
the reader should have in mind that if the domain of @xmath161 is the whole interval @xmath162 , then @xmath189 if and only if @xmath190 , i.e. @xmath191 $ ] .
also , if the multifractal formalism holds at @xmath192 , then @xmath193 let @xmath194))$ ] and consider the function @xmath195 we have @xmath196 .
let @xmath5 be the random weak gibbs measure on @xmath197 obtained as @xmath198 , where @xmath199 is obtained from proposition [ eigen ] .
without changing the random measures @xmath200 and @xmath201 , we can assume @xmath202 .
then , since equation , for any @xmath203 , there exists a unique @xmath204 such that @xmath205 and the mapping @xmath206 is concave and non decreasing .
[ multifractal initial ] for @xmath28-a.e .
@xmath84 , 1 . for all @xmath207 , @xmath208 2 .
the multifractal formalism holds globally for @xmath201 .
furthermore , for all @xmath209 $ ] , one has @xmath210 3 . for all
@xmath211 $ ] , @xmath212\}. \end{aligned}\ ] ] 4 .
for all @xmath209 $ ] , @xmath213 we refer to @xcite for the notions of generalized hausdorff and packing measures @xmath214 and @xmath215 associated with a gauge function @xmath216 , i.e. a function @xmath217 which is non decreasing and satisfies @xmath218 .
[ multifractal initial2 ] for all @xmath209 $ ] such that @xmath219 , for all gauge function @xmath216 , the following zero - infinity laws hold : @xmath220 let us put our result in perspective with respect to the existing literature .
the study achieved in @xcite leads to the multifractal nature of gibbs measures projected on some random cantor sets whose construction assumes a strong separation condition for the pieces of the construction .
about the same time , the multifractal analysis of random gibbs measures and birkhoff averages on random cantor sets and the whole torus were obtained in @xcite ; when the support of the measure is a cantor set , a strong separation condition is assumed as well .
more recently , in @xcite , the multifractal analysis for disintegrations of gibbs measures on @xmath221 was achieved as a consequence of the multifractal analysis of conditional birkhoff averages of random continuous potentials ( not @xmath222 ) .
the approach developed there could , with some effort , be adapted to derive our results on weak gibbs measures if we worked with random fullshift only .
however , as we already said it in the beginning of the introduction , the method can not be extended easily to the random subshift , and our view point will be different . in @xcite ,
the authors start by establishing large deviations results , and then use them to construct by concatenation moran sets of arbitrary large dimension in the level sets @xmath223 ; we will concatenate information provided by random gibbs measures associated with hlder potentials which approximate the continuous potentials associated with the random weak gibbs measure and the random maps generating the attractor @xmath224 .
this will provide us with a very flexible tool from which , for instance , we will deduce the result about the sets @xmath225 . in this sense , our results also complete a part of those obtained in @xcite which , in particular , achieves the multifractal analysis of random gibbs measures on random cantor sets obtained as the repeller of random conformal maps . the multifractal analysis of birkhoff averages on random conformal repellers of @xmath0 expanding maps is studied in @xcite , where the random dynamics is in fact coded by a non random subshift of finite type , and the random potentials that are considered satisfy an equicontinuity property stronger than the one we require .
the sets @xmath226 were studied for deterministic gibbs measures on conformal repellers and for self - similar measures in @xcite .
finally , in @xcite , zero - infinity laws are established for besicovitch subsets of self - similar sets of the line .
this inspired theorem [ multifractal initial2 ] , of which the results in @xcite turn out to be a special case .
also , in @xcite , a zero - infinity law is established for the hausdorff and packing measure of sets of generic points of invariant measures on a conformal repeller .
[ higherdim ] the approach used in this paper can be extended to the higher dimension if the random attractor can be represented as follows : for any @xmath84 , let @xmath227 be closed sets which are the closures of their interiors supposed to be pairwise disjoint .
we assume that 1 .
@xmath228^d$ ] for any @xmath229 .
2 . there exists a measurable @xmath0 diffeomorphism @xmath131^d$ ] .
we denote the inverse of @xmath133 by @xmath134 .
furthermore , @xmath230^d$ ] which is defined as @xmath231 is measurable .
3 . there exists a function @xmath232^d))$ ] such that @xmath233 for any @xmath234 with @xmath229 .
+ here , we say that a measurable function @xmath139 defined from @xmath235 to @xmath183 is in @xmath141^d))$ ] if * @xmath142 , where @xmath236 , * for @xmath28-a.e .
@xmath84 , @xmath144 as @xmath145 , where @xmath237 4 .
equation holds .
now , we can define @xmath238^d),\ \forall v = v_0v_1\cdots v_{n-1}\in\sigma_{\omega , n}\\ x_{\omega}&=&\bigcap_{n\geq 1}\bigcup_{v\in\sigma_{\omega , n}}u_{\omega}^v\\ x_{\omega}&=&\{(\omega , x):\omega\in\omega , x\in x_{\omega}\}. \end{aligned}\ ] ] nevertheless there is a difference in the estimation of the local dimensions of measures .
as will see , in this paper , a building block in our proofs is the comparison of the mass assigned by a random gibbs measure to neighboring basic intervals of the form @xmath239 , in order to control the mass assigned to centered intervals by a random weak gibbs measure , as well as some auxiliary measures obtained by concatenation of pieces of random gibbs measures ( this point of view is fruitful in the study of the discrete inverses of random weak gibbs measures in the companion paper mentioned at the beginning of this introduction ) . in higher dimension
the situation is different in general .
if a strong separation condition is satisfied by the basic sets @xmath239 , there is no much difference with the 1 dimensional case .
otherwise , one can adapt the method used in @xcite for self - conformal measures and under the open set condition , which , for any gibbs measure @xmath240 , consists in controlling the asymptotic behavior of the distance of @xmath240-almost every point @xmath241 to the boundary of the basic set of the @xmath54-th generation containing @xmath241 .
we omit the details .
we end this section with examples illustrating our assumptions on the random attractors considered in this paper . as a first example , one has the fibers of mcmullen - bedford self - affine carpets , and more generally the gatzouras - lalley self - affine carpets @xcite , which naturally illustrate the idea that at a given step of the construction two consecutive intervals @xmath242 and @xmath243 may touch each other . in @xcite , luzia considers a class of expanding maps of the 2-torus of the form @xmath244 that are @xmath245-perturbations of gatzouras - lalley carpets , whose fibers illustrate our purpose with nonlinear maps .
these examples are associated with random fullshift .
let us give a first more explicit example associated with a random subshift and a piecewise linear random maps .
let @xmath246 @xmath247 be the @xmath27-algebra generated by the cylinders @xmath248 $ ] ( @xmath249 , @xmath250 for @xmath251 ) , and @xmath28 the probability measure on @xmath130 defined by @xmath252)=\frac{1}{n_1(n_1 + 1)}\cdot\frac{1}{n_2(n_2 + 1)}\cdot\dots\cdot \frac{1}{n_k(n_k+1)}.\ ] ] also , let @xmath27 be the shift map on @xmath79 .
such a system is ergodic .
it satisfies the conditions we need . for @xmath253 ,
define @xmath254 and @xmath255 the @xmath256-matrix with all entries equal to 1 if @xmath257 or @xmath258 , and the @xmath256-matrix whose @xmath259 first rows have entries equal to 1 and the entries of the @xmath260-th row equal 0 except that @xmath261 .
it it is easy to check that both @xmath29 and @xmath262 are measurable , that @xmath263 , and @xmath29 and @xmath262 define a random subshift , which is not a fullshift . also , the integer @xmath264 is measurable , since for any @xmath265 , @xmath266.\ ] ] notice that both @xmath29 and @xmath48 are unbounded .
then we set @xmath267 for @xmath268 $ ] and for @xmath269 .
in fact , the measure @xmath270 defined above is a special example of a gibbs measure on @xmath271 ( see @xcite ) .
so we can enrich the previous construction by considering any such measure @xmath270 for which @xmath272 .
for the mappings maps @xmath273 , here is a way to provide a non trivial example , which seems to be not covered by the existing literature .
start with a family @xmath274 of random @xmath0 differeomorphisms of @xmath25 $ ] such that at least one @xmath275 is nowhere @xmath276 with positive probability .
assume that there exists a random variable @xmath277 taking values in @xmath278 $ ] and such that @xmath279 } |\varphi'_{s,\omega}(x)|\ge a_0(\omega).\ ] ] let @xmath280 , where @xmath281 is the linear map from @xmath282 onto @xmath25 $ ] .
then , the constant @xmath283 of equation satisfies @xmath284\,\mathrm{d}\mathbb p(\omega).\ ] ] thus , we require that @xmath285\,\mathrm{d}\mathbb p(\omega)>0.\ ] ] this allows some @xmath273 be not uniformly expanding , but ensures expansiveness in the mean .
it is easily seen that the lebesgue measure of @xmath224 is almost surely bounded by @xmath286 for all @xmath51 .
thus , if we strengthen our requirement by assuming that @xmath287 \,\mathrm{d}\mathbb p(\omega)>0,\ ] ] then the lebesgue measure of @xmath224 is 0 almost surely . now let us provide a completely explicit illustration of the last idea ( we will work with a random fullshift for simplicity of the exposition ) .
we take @xmath288 as the fullshift @xmath289 . for any @xmath54-th cylinder @xmath290\subset \omega$ ] we set @xmath291)=\frac{1}{3^n}$ ] .
it is the unique ergodic measure of maximal entropy for the shift map .
let @xmath29 be a random variable depending on @xmath292 only , which is given by @xmath293 the entries of the random transition matrix are always 1 ( we consider the random fullshift ) .
we assume that the map @xmath294 just depends on @xmath292 and @xmath241 .
if @xmath295 , let @xmath296 for @xmath297 and @xmath298 $ ] , @xmath299 $ ] , @xmath300 $ ] and @xmath301 $ ] . in this case , we know that @xmath302 ; notice that the intervals @xmath303 cover the interval @xmath25 $ ] . if @xmath304 , let @xmath305
define @xmath306 and @xmath307 $ ] . in this case
we can choose @xmath308 .
it is easy to check that @xmath309 is not expanding on some interval ; furthermore it is just of class @xmath0 since @xmath310 is nowhere @xmath311-hlder for any @xmath312 .
if @xmath313 , let @xmath296 for @xmath314 and @xmath315 , and @xmath316 $ ] , @xmath317 $ ] , @xmath318 $ ] .
it is easily checked that the left derivative of @xmath309 and the right derivative of @xmath319 do not coincide , so the dynamics is not the restriction of a random conformal map . in this case
we can choose @xmath320 . also , @xmath321
\,\mathrm{d}\mathbb p(\omega)=\frac{\log 21-\log 16}{3}>0,\ ] ] so that all the conditions hold .
this section prepares the proofs of our main results .
sections 3.1 and 3.2 present basics facts about random weak gibbs and random gibbs measures .
section 3.3.1 provides an approximation of @xmath322 by a family @xmath323 of random hlder potentials .
section 3.3.2 derives some related properties of the associated pressure functions , which yield the variational formulas appearing in theorem [ multifractal initial ] .
section 3.3 presents properties related to the random gibbs measures associated with the couple @xmath324 , which will be used as building blocks in the concatenation of measures used in the proof of the main parts of theorem [ multifractal initial ] and of theorem [ multifractal initial2 ] ( section 4 ) .
fix a potential @xmath110 ( here @xmath113 may not be 0 ) . since @xmath325 as @xmath326 and @xmath327 is bounded in @xmath328 norm , using maker s ergodic theorem @xcite , we can get @xmath329 where @xmath330 .
due to ( [ int ] ) and the ergodic theorem , setting @xmath331 , for any positive sequence @xmath332 such that @xmath333 , @xmath28-almost surely we have @xmath334 a family @xmath335 of measurable maps satisfying @xmath336 for all @xmath337 and @xmath338 is called an extension .
we say that it is measurable if the mapping @xmath339 is measurable for all @xmath340 .
let @xmath341 be an extension and @xmath110 .
then for @xmath337 @xmath342 is called @xmath54-th partition function of @xmath103 in @xmath49 with respect to @xmath343 .
due to the assumption @xmath344 , using the same method as in @xcite , it is easy to prove the following lemma .
[ converge ] let @xmath343 be any extension and @xmath110 .
then @xmath345 for @xmath28-almost every @xmath84 .
this limit is independent of @xmath343 .
furthermore , the variation principle holds , which means , @xmath346 now let @xmath347 where @xmath348 is defined as in proposition [ eigen ] .
the following lemma is direct when the potential @xmath103 possesses bounded distorsions so that the ruelle - perron - frobenious theorem holds for the operator @xmath349 . for general potentials in @xmath104
we need a proof .
[ lambda ] one has @xmath350 for @xmath28-almost every @xmath84 . in the next proof , as well as in the rest of the paper , we will use the letter @xmath48 to denote the levels of the function @xmath351 . keeping this in mind
should prevent from some confusion . for @xmath352 ,
let @xmath353 .
fix @xmath48 large enough so that @xmath354 . for each @xmath84 ,
let @xmath355 be the @xmath356-th return time of @xmath49 to the set @xmath357 , i.e. @xmath358 and for @xmath359 , @xmath360 . from ergodic theorem
we get @xmath361 for @xmath28-almost every @xmath116
. then @xmath362 for @xmath28-almost every @xmath116 , which implies that @xmath363 .
now we claim that for any @xmath364 we have @xmath365 the second inequality uses the fact that we work with a subshift as well as .
we just prove the first inequality : for @xmath54 large enough so that @xmath366 , @xmath367 by using the topological mixing property and preserving for each @xmath368 only one path of length @xmath369 from @xmath370 to @xmath371 , the inequality follows from ( [ vn ] ) , @xmath372 and ( [ phi ] ) .
now , since @xmath373 , we can easily get the result from lemma [ converge ] and the fact that @xmath372 . for each @xmath374 ,
let @xmath375 then @xmath376 for @xmath28-almost every @xmath116 . from @xmath372 , and lemma [ lambda ]
we can get that @xmath377 @xmath28-almost surely .
recall that by proposition [ eigen ] , for @xmath28-a.e @xmath116 , the measures @xmath200 satisfy @xmath378 . [ wgibbs ] for @xmath28-a.e @xmath84 , for any @xmath340 , for all @xmath379 , one has @xmath380_{\omega}}s_n\phi(\omega , v))}\leq \widetilde \mu_{\omega}([v]_{\omega})\leq \frac{1}{\lambda(\omega , n)}\exp{(\sup_{{\underline{v}}\in[v]_{\omega}}s_n\phi(\omega,{\underline{v}}))},\ ] ] so that @xmath381_{\omega})}{\exp(s_n\phi(\omega,{\underline{v}})-\log(\lambda(\omega ,
n)))}\leq \exp(\epsilon_n n)\ ] ] for any @xmath382_{\omega}$ ] , where @xmath383 does not depend on @xmath384 and tends to 0 as @xmath108 .
let us deal first with the case @xmath385 .
fix @xmath386 . for any @xmath387
, there exists @xmath388 such that @xmath389 . due to proposition [ eigen ]
, we have @xmath390_{\omega})=\frac{1}{\lambda(\omega , m(\omega))}\int \mathcal{l}_{\phi}^{\omega , m(\omega ) } 1_{[iwj]_{\omega } } \ , \mathrm{d } \widetilde\mu_{\sigma^{m(\omega)}\omega},\ ] ] where @xmath391 .
this implies @xmath390)\geq \frac{\inf_{{\underline{w}}\in [ iwj]_{\omega}}\exp(s_{m(\omega)}\phi(\omega,{\underline{w}}))}{\lambda(\omega , m(\omega ) ) } \widetilde\mu_{\sigma^{m(\omega)}\omega}([j]_{\sigma^{m(\omega)}\omega}).\ ] ] then @xmath392)\geq d(\omega)$ ] follows after summing over @xmath387 .
the upper bound @xmath392)\le 1 $ ] is obvious .
the general case is achieved similarly : if @xmath393 , for each @xmath394 , there exists @xmath395 such that @xmath396 .
one has @xmath397_{\omega})=\frac{1}{\lambda(\omega , n ) \lambda(\sigma^n\omega , m(\sigma^n\omega))}\int \mathcal{l}_{\phi}^{\omega , n+m(\sigma^n\omega ) } 1_{[vwj]_{\omega } } \,\mathrm{d } \widetilde\mu_{\sigma^{n+m(\sigma^n\omega)}\omega},\ ] ] from which we get @xmath397_{\omega})\ge \frac{1}{\lambda(\omega , n ) } \cdot d(\sigma^n\omega ) \exp{(\inf_{{\underline{v}}\in[v]_{\omega}}s_n\phi(\omega , v ) ) } \widetilde\mu_{\sigma^{n+m(\sigma^n\omega)}\omega}([j]_{\sigma^{n+m(\sigma^n\omega)}\omega}).\ ] ] then , taking the sum over @xmath394 we get @xmath398_{\omega})\geq\frac{d(\sigma^n\omega)}{\lambda(\omega , n)}\exp{(\inf_{{\underline{v}}\in[v]_{\omega}}s_n\phi(\omega , v))}.\ ] ] the inequality @xmath399_{\omega})\leq \frac{1}{\lambda(\omega , n)}\exp{(\sup_{{\underline{v}}\in[v]_{\omega}}s_n\phi(\omega,{\underline{v}}))}$ ] is direct from the equality @xmath400_\omega)=\frac{1}{\lambda(\omega , n ) } \int \mathcal{l}_{\phi}^{\omega , n } 1_{[v]_{\omega}}\ , \mathrm{d } \widetilde\mu_{\sigma^{n}\omega}.\ ] ] finally we conclude with ( [ vn ] ) and @xmath377 . for any @xmath401))$ ] and any @xmath402 , let @xmath403 from the lagrange s finite - increment theorem , distortions and proposition [ wgibbs ] , using standard estimates we can get the following proposition . [ for n ] for @xmath28-almost every @xmath116 , there are positive sequences @xmath404 and @xmath405 , that we also denote as @xmath406 and @xmath407 , converging to 0 as @xmath408 , such that for all @xmath409 , for all @xmath410 , we have : 1 . for all
@xmath411 @xmath412 hence for all @xmath382_{\omega}$ ] , @xmath413 consequently , for all @xmath414 : @xmath415 2 . for all @xmath416_{\omega}$ ] , @xmath417_{\omega})\leq \exp(s_n\phi(\omega,{\underline{v}})+n\epsilon(\phi , n)),\ ] ] hence for all @xmath402 , @xmath418 as well as @xmath419 if @xmath200 is atomless .
random gibbs measures are associated with random hlder continuous potentials .
we say that a function @xmath103 is a random hlder potential if @xmath103 is measurable from @xmath73 to @xmath183 , @xmath420 and there exists @xmath421 $ ] such that @xmath422 where the random variable @xmath423 is such that @xmath424 .
a random hlder continuous potential is obviously in @xmath104 .
assume that @xmath247 is a countably generated @xmath27-algebra , @xmath93 is a topological mixing subshift of finite type and @xmath103 a random hlder potential . for @xmath28-almost
every @xmath84 , there exists some random variables @xmath425 , @xmath426 , a function @xmath427 and a measure @xmath428 with disintegrations @xmath429 satisfying @xmath430 and such that @xmath431 let @xmath432 be given by @xmath433 and set @xmath434 .
then @xmath435 , and for @xmath28-almost every @xmath116 , for all @xmath436 , and for all @xmath416_{\omega}$ ] @xmath437_{\omega})}{\exp(\sum_{i=0}^{n-1}\phi(f^i(\omega,{\underline{v}}))-\log\lambda^{\phi}(\sigma^{n-1}\omega)\cdot\dots \cdot\lambda^{\phi}(\omega))}\leq c^{\phi}.\ ] ] the family of measures @xmath438 is called a random ( or relative ) gibbs measure for the potential @xmath103 .
moreover , @xmath439 is the unique maximizing @xmath93-invariant probability measure in the variational principle , i.e. such that @xmath440 each time we need to refer to the function @xmath103 , we denote the measures @xmath439 and @xmath441 as @xmath442 and @xmath443 , and denote @xmath444 as @xmath445 .
we can also define a random gibbs measure on the random attractor @xmath224 by setting @xmath446 given a random hlder potential @xmath103 , from we can define the normalized potential @xmath447 , which satisfies @xmath448 for @xmath28-almost every @xmath116 .
this implies that @xmath449 for @xmath28-almost every @xmath116 .
also , we have the following fact : [ control phi ] suppose that @xmath103 is a random hlder potential . if @xmath202 , there exist some @xmath450 such that for @xmath28-almost every @xmath116 , there exists @xmath451 such that for any @xmath452 and any @xmath393 , one has @xmath453_{\omega}}s_{n}\phi(\omega,\underline{v})\leq -n\varpi.\ ] ] as a consequence , @xmath201 is atomless .
if we need to refer explicitly to @xmath103 , we will use the notations @xmath454 and @xmath455 instead of @xmath451 and @xmath456 .
the main idea of the proof is from @xcite . since @xmath202
, we have @xmath457 .
we claim that @xmath458 .
let @xmath48 large enough such that @xmath459 .
for any @xmath84 such that @xmath460 and @xmath461 we have @xmath462 , hence @xmath463 for any @xmath464 and @xmath465 for any probability measure @xmath466 on @xmath86 . since , moreover , we have @xmath467 , we conclude that @xmath468 let @xmath469 .
if the conclusion of the proposition does not hold , there exists a subsequence @xmath470 such that @xmath471_{\omega}}s_{n_k}\phi(\omega,{\underline{v}})}{n_k } > -\varpi\}\geq 1 $ ] , hence for any @xmath472 , @xmath473 however , @xmath474 since @xmath475 , letting @xmath476 tend to @xmath2 we get a contradiction . we mainly introduce objects and related properties which will be used in the following sections .
also , we explain the variational formulas appearing in the statement of theorem [ multifractal initial ] .
now we approximate the potentials @xmath103 and @xmath477 associated with @xmath478 and @xmath479 by more regular potentials : for any @xmath480 , for any @xmath116 for any @xmath481_{\omega}\subset \sigma_{\omega}$ ] with @xmath482 define @xmath483_{\omega}\}+\min\{\phi(\omega,{\underline{w}}),{\underline{w}}\in [ v]_{\omega}\}}{2},\ ] ] @xmath484_{\omega}\}+\min\{\psi(\omega,{\underline{w}}),{\underline{w}}\in [ v]_{\omega}\}}{2}.\ ] ] these functions @xmath485 and @xmath486 are piecewise constant with respect to the second variable .
they are random hlder continuous potentials .
if we take @xmath487 then @xmath488 furthermore , @xmath489 and the right hand side is integrable since @xmath490 . also , since for @xmath28-almost every @xmath49 we have @xmath325 as @xmath408 , and @xmath491 , we have @xmath492 uniformly as @xmath493 for @xmath28-almost every @xmath49 .
the same property holds for @xmath486 and @xmath477 .
consequently , without loss of generality we can also assume that @xmath494 since @xmath495 converges to @xmath113 as @xmath496 tends to @xmath497 .
due to our assumptions on @xmath322 and the definition of @xmath500 , we have @xmath501 , hence for the same reason as for @xmath322 , for any @xmath203 , for any @xmath502 , there exists a unique @xmath503 such that @xmath504 and the function @xmath505 is concave and non - decreasing .
also , the function @xmath505 is differentiable since for hlder potentials the associated random gibbs measure is the unique invariant measure that maximizes the variation principle ( see @xcite . )
[ converge t ] for any @xmath203 , one has that @xmath506 as @xmath493 . at first
, we recall that for any @xmath110 one has @xmath507 also , for any @xmath203 , we have @xmath508 .
thus @xmath509\ , \mathrm{d}\rho\right ) \leq 0,\ ] ] and @xmath510\ , \mathrm{d}\rho\right ) \geq 0.\ ] ] the inequality implies that for any @xmath511 , there exists a measure @xmath512 such that @xmath513\ , \mathrm{d}\rho \leq \varepsilon.\ ] ] then @xmath514 and @xmath515 since @xmath516 for @xmath496 large enough . letting @xmath493 , from the arbitrariness of @xmath517 we get @xmath518 using similarly we can get @xmath519 finally @xmath520 [ convergent t * ] let @xmath521 be a concave function .
suppose that @xmath522 is a sequence of differentiable concave functions from @xmath162 to @xmath162 which converges pointwise to @xmath523 .
then @xmath524 converges pointwise to @xmath525 over the interior of the domain of @xmath525 .
let @xmath526 be an interior point of @xmath527 .
let @xmath528 be the unique point such that @xmath529 $ ] , and @xmath530 . by ( * ?
* proposition 2.5(i ) ) , there exists a sequence @xmath531 such that for @xmath496 large enough one has @xmath532 . without loss of generality
we can assume that this sequence converges to @xmath533 or diverges to @xmath534 or @xmath2 .
suppose first that it converges to @xmath535 .
if @xmath536 then we are done since @xmath522 converges uniformly on compact sets .
suppose that @xmath537 and @xmath538 . using the uniform convergence of @xmath522 in a compact neighborhood of @xmath539 $ ] and the inequality @xmath540 ( @xmath541 is concave )
, we can get @xmath542 . on the other hand ,
@xmath206 being concave we have @xmath543 and @xmath544 .
this implies that @xmath545 hence @xmath546 .
the case @xmath537 and @xmath547 is similar .
now suppose that @xmath531 diverges to @xmath2 ( the case where it diverges to @xmath534 is similar ) .
if @xmath523 is affine over @xmath548 with slope @xmath526 , @xmath526 is not an interior point of @xmath527 .
consequently , there exists @xmath549 and @xmath550 such that @xmath551 , and @xmath552 for all @xmath553 . on the other hand ,
since @xmath554 is non increasing for all @xmath496 , for @xmath496 large enough we have @xmath555 for all @xmath556 $ ] .
since @xmath531 diverges to @xmath2 , this contradicts the convergence of @xmath522 to @xmath523 .
[ pro : for t ] for any @xmath557 we have @xmath558 furthermore , for any @xmath209 $ ] we have @xmath559 the proof of is just use the fact of the variation principle , see . regarding the equation , on the one hand , for any @xmath560 , @xmath561 on the other hand , for any @xmath562 , by the proof of lemma [ convergent t * ] , there exists @xmath496 large enough and @xmath563 such that @xmath564 and @xmath565
. then there exists @xmath566 such that @xmath567 and @xmath568 this implies that there exists @xmath566 such that @xmath569 and @xmath570 lemma [ convergent t * ] tells us that @xmath571 as @xmath493 and @xmath572 is compact for the weak * topology . thus , there exists a limit point @xmath573 of @xmath574 in @xmath575 such that @xmath576 since the entropy map is upper semi - continuous and @xmath324 converges uniformly to @xmath322 . finally , we get @xmath577 the case @xmath578 now follows by approximating @xmath192 by a sequence @xmath579 of elements of @xmath580 and for each @xmath356 picking @xmath581 which realizes @xmath582 . then , since @xmath583 is continuous at @xmath192 ( it is lower semi - continuous as a concave function and upper semi - continuous as a legendre transform ) , any limit point of @xmath584 is such that @xmath585 and @xmath586 .
it exists since @xmath572 is compact in the weak * topology ( see @xcite ) . in this quite technical subsection , we prepare the `` concatenation of random gibbs measures '' approach that will be used in the next sections to construct auxiliary measures with nice properties .
we also show an almost everywhere almost doubling property for the random gibbs measures on the random attractor @xmath224 .
let @xmath587 be a dense and countable subset of @xmath580 .
let @xmath588 be an increasing sequence of finite sets such that @xmath589 .
fix a sequence @xmath590 decreasing to 0 as @xmath493 . due to lemma
[ converge t ] and the proof of lemma [ convergent t * ] , for any @xmath591 there exists @xmath592 large enough such that for any @xmath593 , there exists @xmath594 such that the following properties hold : 1 .
@xmath595 , @xmath596 .
2 . @xmath597 and @xmath598 we can also assume that @xmath599 for each @xmath600 .
we set @xmath601 for any @xmath602 , we define @xmath603 for the convenience of writing , we will denote @xmath604 by @xmath605 , so that @xmath606 we also define @xmath607 . recall proposition [ for n ] . for each @xmath608 , for any @xmath609 , there exist positive integers @xmath610 ( large enough ) and @xmath611 such that there exist a set @xmath612 and a sequence @xmath613 decreasing to @xmath1 as @xmath614 such that : @xmath615 , and for any @xmath616 , one has : * @xmath617 , * for any @xmath51 , @xmath618 and @xmath619 where we have used egorov s theorem ; * for all @xmath620 , for @xmath621 @xmath622 @xmath623_{\omega } } s_n\psi(\omega,{\underline{v}})\leq ( -n\varpi_{\psi } ) , \ \forall v\in \sigma_{\omega , n},\ ] ] and @xmath624 where we have applied ergodic theorem and egorov s theorem . * for any @xmath602 , the measure @xmath625 is well defined , and for any @xmath626 @xmath627 @xmath623_{\sigma^{m}\omega } } s_n \lambda_{i , q}(\sigma^{m}\omega,{\underline{v}})\leq ( -n\varpi_{i}).\ ] ] let @xmath628 be the @xmath629-th return time of the point @xmath49 to the set @xmath612 under the map @xmath27 , that is @xmath630 and for any @xmath631 and @xmath632 , @xmath633 then for any @xmath600 , from the ergodic theorem , for @xmath28-almost every @xmath49 @xmath634 consequently , @xmath635 since @xmath636 is countable , there exists @xmath637 of full probability such that for all @xmath638 , for any @xmath591 , we have @xmath639 hence @xmath640 given @xmath641 , let @xmath642 for @xmath643 , define @xmath644 , where @xmath645 is the smallest @xmath629 such that the following property holds : @xmath646 it is easily seen that @xmath647 now we prove an almost everywhere almost doubling property for the gibbs measures @xmath648 . for @xmath649 ,
we denote by @xmath650 and @xmath651 the two intervals of the @xmath54-th generation of the construction of @xmath224 which are neighboring @xmath652 , whenever @xmath652 is neither the leftmost nor the rightmost of the whole collection , and with the convention that @xmath651 is on the left of @xmath652 .
we say that @xmath653 holds if @xmath654 , and @xmath655 or @xmath656 .
let @xmath657 [ control ] for all @xmath502 , for all @xmath616 , for all @xmath602 , we have @xmath658 for any @xmath384 and @xmath659 such that @xmath660 and @xmath661 , by construction of @xmath662 one has @xmath663 we will use this fact to estimate the measure of @xmath664 . notice that for any @xmath665 , there are at most two @xmath659 such that @xmath661 and @xmath666 holds .
consequently , @xmath667 since @xmath611 and @xmath668 we get @xmath669 by borel - cantelli s lemma we get @xmath670 . for any @xmath511 , @xmath671 , and @xmath672
we now define the following sets : @xmath673 @xmath674 and then @xmath675 [ initial ] for all @xmath502 , for any @xmath511 , for all @xmath616 , for all @xmath602 , the singularity set @xmath676 has full @xmath662-measure .
fix @xmath511 .
let @xmath677 we have @xmath678 since @xmath679 is in fact not only differentiable , but analytic @xcite , we have @xmath680 uniformly in @xmath602 .
thus , there exists @xmath681 such that for @xmath682 small enough , for all @xmath602 , we have @xmath683 consider such an @xmath682 in @xmath684 $ ] .
we have @xmath685 consequently , @xmath686 which by the borel - cantelli lemma yields the desired conclusion since @xmath517 is arbitrary .
now we can collect the following facts .
[ control initial ] lemma [ control ] and [ initial ] imply that for all @xmath502 , for any @xmath687 , for all @xmath616 , there exists an integer @xmath688 such that for any @xmath602 , there exists @xmath689 such that 1 .
@xmath690 , 2 .
@xmath691 , 3 .
@xmath692 , 4 .
@xmath693 for any @xmath694 , 5 .
[ approximate 5 ] for any @xmath695 , for any @xmath696 with @xmath694 such that @xmath382_{\sigma^{m(i)}\omega}$ ] , one has @xmath697 and @xmath698 .
furthermore , for any @xmath699 , there exists ( in fact for all ) @xmath700_{\sigma^{m(i)}\omega}$ ] such that @xmath701 @xmath702 and @xmath703 in fact with a suitable change of @xmath704 ( take it as @xmath705 ) , we can get the following additional properties from , , and above : @xmath706 @xmath707 and @xmath708 [ modify ] we can change @xmath612 to @xmath709 a bit smaller such that @xmath710 and there exist @xmath711 and @xmath712 such that for any @xmath713 , @xmath714 and @xmath715 and the properties listed in facts [ control initial ] hold .
we define @xmath716 as being the @xmath629-th return time to the set @xmath717 for the point @xmath49 .
since @xmath636 is countable , there exists @xmath718 of full probability such that for all @xmath719 , for any @xmath591 , we have @xmath720 hence @xmath721
this section consists of three subsections . in the first one we obtain
the sharp lower bound for the lower @xmath4-spectrum of @xmath201 . next ,
in the second subsection , we prove the validity of the mutifractal formalism ( see theorem [ multifractal initial](2 ) ) .
due to , and lemma [ pro : for t ] , we just need to prove @xmath722 . there
, our approach to construct suitable auxiliary measures already prepares the material used to establish in the third subsection the refinements gathered in theorem [ multifractal initial](3)(4 ) and theorem [ multifractal initial2 ] . [ [ lower - bound - for - tau_mu_omega - and - upper - bound - for - tau_mu_omega ] ] lower bound for @xmath723 and upper bound for @xmath724 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1 . for all @xmath726 the weak gibbs measure @xmath727 are defined ; 2 .
for all @xmath728 the conclusions of proposition [ for n ] hold all the potentials @xmath729 , @xmath726 ; 3 . for all @xmath54 large enough , for all @xmath393 , for all @xmath382_\omega$ ]
, we will establish the lower bound @xmath735 for all @xmath728 and @xmath726 .
since @xmath587 is dense and both @xmath723 and @xmath206 are continuous , this will yield @xmath736 for all @xmath728 . by using the multifractal formalism
, this immediately yields the desired upper bound @xmath583 for @xmath724 and the various spectra we consider for @xmath201 .
the equality @xmath737 then follows from standard considerations in large deviations theory
. fix @xmath728 .
let @xmath738 and consider @xmath739 , a packing of @xmath740 by disjoint balls @xmath8 with the center @xmath169 and radius @xmath9 .
for each ball @xmath8 , choose @xmath741 and @xmath742 such that @xmath743 and @xmath744 , but @xmath745 . by removing a set of probability 0 from @xmath725 if necessary , for any @xmath746_\omega$ ]
, we have @xmath747 where we have used ergodic theorem .
thus @xmath748 for @xmath9 small enough . on the other hand , for @xmath9 small enough , for any @xmath746_\omega$ ] we have @xmath749so @xmath750 .
thus , for @xmath9 small enough , independently on @xmath751 , if @xmath742 and @xmath746_\omega$ ] we have @xmath752 for each @xmath754 , one has @xmath755 , so for any @xmath746_\omega$ ] @xmath756 where we have applied proposition [ for n](2 ) to the potential @xmath729 as well as proposition [ for n](1 ) , the fact that @xmath757 and .
it follows that @xmath758 , and this bound does not depend on the choice of the packing @xmath759 . letting @xmath760 yields @xmath761 .
* case @xmath762 : * define @xmath763 @xmath764 @xmath765 then @xmath766 is a partition of @xmath25 $ ] .
define @xmath767 and @xmath768 .
then , from we now that for some positive constants @xmath769 and @xmath770 , for @xmath9 small enough , we have @xmath771 . for any @xmath772 , @xmath773 meets at most @xmath774 many balls of @xmath8 and for any @xmath8 , @xmath8 meets at most two intervals of @xmath775 with @xmath776 .
consequently , since @xmath777 , we have @xmath778 using the same argument as for @xmath779 , we can know get that @xmath780 so that @xmath781 independently on @xmath759 , where we used the fact that @xmath766 is a partition of @xmath25 $ ] . letting @xmath760 yields @xmath761 .
recall facts [ control initial ] and facts [ modify ] derived at the end of section [ subsection : simultaneous control of auxiliary random measures ] . for any @xmath782 , for any @xmath209 $ ] , for any sequence @xmath783 with @xmath784 , such that @xmath785 , and consequently @xmath786 by continuity of @xmath583 , we will construct a probability measure @xmath787 supported on a set @xmath788 such that * we only deal with points @xmath49 in the set @xmath797 of @xmath270-probability 1 for which the sequence @xmath798 is well defined ; * the sequence @xmath799 is fixed ; * for each @xmath713 the sequence @xmath800 is well defined ; * we denote the properties listed in fact [ control initial ] by @xmath801 .
more precisely , we denote the five items by @xmath802 to @xmath803 . for any @xmath782 , recall that @xmath805 is the smallest @xmath409 such that @xmath806 .
. facts [ control initial ] and facts [ modify ] tell us that for any @xmath808 , there exists @xmath809 such that @xmath810 and a set @xmath811 such that @xmath812 hold .
* @xmath814 .
* @xmath815 , * @xmath816 , * for any @xmath629 such that the return time @xmath817 satisfies @xmath818 , one also has @xmath819 + let @xmath820 be the smallest @xmath629 such that @xmath821 .
( b1 ) at ( -0.1,1 ) @xmath49 ; ( -0.1,0.7 ) ( -0.1,0 ) ; ( 0,0 )
( 1,0 ) ; ( b2 ) at ( 0.5,0.5 ) @xmath54 ; ( b3 ) at ( 1,-1 ) @xmath826 ; ( 1.1,-0.7 )
( 1.1,0 ) ; ( 1.15,0 )
( 2.06,0 ) ; ( b4 ) at ( 1.6,0.5 ) @xmath827 ; at ( 2.7,0 ) [ circle , draw = blue!50 ] @xmath828 ; ( 3.3,0 )
( 6,0 ) ; ( b5 ) at ( 4.7,1 ) long time ; ( 4.7,0.7 ) ( 4.7,0.1 ) ; ( 6.1,0 ) ( 15,0 ) ; at ( 1.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; at ( 3.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; ( 2.7,0 ) ( 5,0 ) ; ( b6 ) at ( 3.8,-0.7 ) @xmath829 ; at ( 6.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; at ( 8.5,0 ) [ circle , draw = red!50,fill= red!30 ] ; ( 10.3,0.1 ) ( 10.3,-0.1 ) ; ( 2.7,0 )
( 10.3,0 ) ; ( b7 ) at ( 6.5,0.7 ) @xmath830 ; at ( 10.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; at ( 12.8,0 ) [ circle , draw = red!50,fill= red!30 ] ; ( 12.8,0.7 ) ( 12.8,0.2 ) ; ( b8 ) at ( 12.8,1 ) @xmath831 ; at ( 14.2,0 ) [ circle , draw = red!50,fill= red!30 ] ; also , set @xmath835 fix @xmath836 for any @xmath808 , we define : @xmath837 and @xmath838 ( recall that the meaning of @xmath839 is specified at the beginning of section [ subsection : random subshift , relativized entropy , topological pressure and weak gibbs measures ] ) .
suppose that @xmath840 have been chosen .
define @xmath841 and @xmath842 like in the case @xmath843 , for any @xmath844 , there exists @xmath845 with @xmath846 as well as a set @xmath847 such that property @xmath848 holds . *
@xmath850 , * @xmath851 * @xmath852 + the above two items ensure that we do not need to wait a long relative time to go to the next step . * for any @xmath629 with @xmath853 one
has @xmath854 let @xmath855 be the smallest @xmath629 such that @xmath856 [ growth speed ] by construction , we can take @xmath861 as big as we want ( in the construction @xmath850 ) .
this implies that we can get the sequence @xmath862 increasing as fast as we want .
we can also impose that @xmath863 and @xmath864 .
thus , the speed we fix for the growth of @xmath865 directly impacts the growth speed of @xmath866 .
( b3 ) at ( -0.1,0 ) @xmath867 ; ( 1.2,0 ) ( 2,0 ) ; ( b4 ) at ( 1.6,1 ) @xmath868 ; at ( 2.7,0 ) [ circle , draw = blue!50,fill= green!30 ] @xmath869 ; ( 3.3,0 ) ( 6,0 ) ; ( b5 ) at ( 4.7,1 ) long time ; ( 4.7,0.7 ) ( 4.7,0.1 ) ; ( 6.1,0 ) ( 15,0 ) ; at ( 1.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; at ( 3.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; ( 2.7,0 ) ( 5,0 ) ; ( b6 ) at ( 3.8,-0.7 ) @xmath861 ; at ( 6.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; at ( 8.5,0 ) [ circle , draw = red!50,fill= red!30 ] ; ( 10.3,0.1 ) ( 10.3,-0.1 ) ; ( 2.7,0 ) ( 10.3,0 ) ; ( b7 ) at ( 6.5,0.7 ) @xmath870 ; at ( 10.7,0 ) [ circle , draw = red!50,fill= red!30 ] ; at ( 12.8,0 ) [ circle , draw = red!50,fill= red!30 ] ; ( 12.8,0.7 ) ( 12.8,0.2 ) ; ( b8 ) at ( 12.8,1 ) @xmath871 ; at ( 14.2,0 ) [ circle , draw = red!50,fill= red!30 ] ; for any @xmath209 $ ] , there exists @xmath876 , such that @xmath877 and @xmath878 .
moreover , if @xmath562 , @xmath879 converges to @xmath880 directly from lemma [ convergent t * ] .
if @xmath881 , again due to lemma [ convergent t * ] , we can choose @xmath882 to be piecewise constant to make sure that @xmath883 tends to 0 as @xmath884 , so that @xmath879 converges to @xmath880 as well .
we fix such a sequence and suppose that the correspondence @xmath476 are @xmath885 . * first estimate .
* for any @xmath888 , for any @xmath889 , for any @xmath890_{\omega}$ ] for @xmath621 we have ( remembering the notations introduced at the beginning of section [ subsection : simultaneous control of auxiliary random measures ] and the fact that by construction we have @xmath891 ) : @xmath892 also , since @xmath893 , and @xmath894 implies @xmath895 , we have @xmath896 one estimates the term invoking @xmath897 similarly
. then , recalling that @xmath898 and by construction @xmath899 and @xmath900 , we get @xmath901 for @xmath496 and @xmath356 , where in the last inequality we have used the fact that @xmath902 and @xmath903 .
we can now estimate the local dimension of @xmath201 .
fix @xmath790 .
if @xmath9 is small enough , we can choose the largest @xmath496 , then the largest @xmath911 , with @xmath904 , such that the following property holds : there exists @xmath912 satisfying @xmath913 and @xmath914 and @xmath915 are the neighboring intervals of @xmath916 , then @xmath917 , @xmath918 are larger than @xmath919 . then by we have @xmath920 and @xmath921 .
so there exists @xmath922 or @xmath923 such that @xmath924 now , using estimates similar to those leading to with @xmath477 replaced by @xmath103 we get that for any @xmath925_{\omega}$ ] , @xmath926 consequently , using , @xmath927 let @xmath928 and @xmath929
. then @xmath930 * if @xmath932 * , there exists @xmath933 such that @xmath934 , @xmath935 and @xmath936 observe that @xmath937 and @xmath938 ( where @xmath939 is defined similarly as @xmath940 ) , so from and we can get @xmath941 consequently , @xmath942 it remains to prove that @xmath953 .
this is easier than for the @xmath16 since we just have to choose the smallest @xmath496 and then the smallest @xmath911 with @xmath954 , such that there exists @xmath955 and @xmath956 for which @xmath913 and @xmath957 .
then @xmath958 if @xmath964 , then there exists @xmath965 , and @xmath966 with @xmath967 and @xmath968 , so @xmath969 in both cases we have @xmath970 finally we have @xmath971 which yields @xmath972 the inclusion @xmath973 is established
. next we estimate @xmath974 from below . for any @xmath975 , with @xmath976 ,
define @xmath977 then inductively , for any @xmath955 and @xmath978 , so that @xmath979 , define : @xmath980 we can extend @xmath787 in a unique way to a probability measure on the @xmath27-algebra generated by @xmath981 .
this measure is supported on @xmath788 .
local]in fact , if in the construction the sequence @xmath866 is replaced by another one growing faster ( with the effect to modify @xmath986 ) ( for example @xmath987 ) , for all @xmath988 , @xmath989 @xmath990 this property will be used in the next subsection . we have @xmath991 where for the first inequality we have used the same estimates as to get .
also , @xmath992 if we assume that @xmath987 .
thus equation is established . to get , observe that @xmath993 if we assume that @xmath987 .
on the other hand , @xmath994 [ [ proofs - of - theoremmultifractal - initial34-and - theoremmultifractal - initial2 ] ] proofs of theorem [ multifractal initial](3)(4 ) and theorem [ multifractal initial2 ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ for the packing dimension , we choose three sequences @xmath882 , @xmath995 and @xmath1009 in @xmath996 in such a way that @xmath997 , @xmath998 , @xmath1010 , @xmath1011 and @xmath1012\}.$ ] then we take @xmath1013 , @xmath1014 and @xmath1015 .
suppose @xmath209 $ ] and @xmath1023 .
the estimates used in the proof of and the proof of remark [ control local ] can be used to get a positive sequence @xmath1024 decreasing to 0 and a constant @xmath1025 such that , independently on @xmath1026 , for all @xmath988 , for @xmath496 large enough , if @xmath1027 , then @xmath1028 since @xmath1029 , we can find @xmath1030 such that @xmath1031 as @xmath493 , @xmath1032 for @xmath496 large enough , @xmath1033 as @xmath493 , and @xmath1034 is ultimately non increasing . for any gauge function @xmath216 such that @xmath1035 , there exists a positive sequence @xmath1036 such that both @xmath1037 and @xmath1038 decrease to 0 as @xmath9 decreases to 0 and @xmath1039 due to for @xmath496 large enough , for any @xmath9 such that @xmath1040 , for any @xmath790 , @xmath1041 notice that @xmath1042 .
so , if we can impose @xmath1043 , we will have @xmath1044 hence @xmath1045 then , for any positive real number @xmath1046 , this will yield @xmath1047 , and letting @xmath1048 , @xmath1049 so @xmath1050 .
finally , suppose that @xmath216 is a gauge function such that @xmath1054 .
there exist @xmath1055 , and @xmath1056 such that @xmath1057 $ ] and @xmath1058 decrease to 0 as @xmath1059 tends to @xmath2 , and @xmath1060 using the same approach as above , we can choose @xmath1061 such that @xmath997 , @xmath1062 converges slowly to @xmath880 from above , and in the construction of @xmath1063 , @xmath1051 tends fast enough to @xmath2 so that , for some @xmath1064 , for all @xmath1065 , for any @xmath1066 , @xmath1067 now , let @xmath1068 be of positive @xmath1003-measure . for any given @xmath1046 , take @xmath1069 such that @xmath1070 consider the following family of closed balls @xmath1071 which is a covering of @xmath262 .
due to besicovitch covering theorem , we can extract an at most countable subfamily of pairwise disjoint balls @xmath1072 such that @xmath1073 .
this family is a @xmath1074-packing of @xmath262 , and ( @xmath1075 stands for the prepacking measure associated with @xmath216 ) @xmath1076 as @xmath1077 when @xmath1048 , we can conclude that @xmath1078 . then , since any at most countable covering of @xmath788 must contain a set @xmath262 of positive @xmath1003-measure , we finally get @xmath1079 , so @xmath1080 .
feng , k .- s . lau and j. wu , ergodic limits on the conformal repellers , _ adv .
_ , 169 , 58 - 91 ( 2002 ) d
feng and e. olivier , multifractal analysis of weak gibbs measures and phase transition pplication to some bernoulli convolutions , _ ergod .
@xmath1081 dynam .
_ , * 23 * ( 2003 ) , 17511784 .
u. frisch and g. parisi , fully developed turbulence and intermittency in turbulence , and predictability in geophysical fluid dynamics and climate dynamics , _ international school of physics _ , * 41 * ( 1985 ) , 8488 .
y. kifer , fractals via random iterated function systems and random geometric constructions , in _ fractal geometry and stochastics ( finsterbergen , 1994 ) , progr .
probab . , 37 _ , birkhuser , basel , ( 1995 ) 145164 .
y. kifer , on the topological pressure for random bundle transformations , in _
topology , ergodic theory , real algebraic geometry _ ,
ser . 2 , 202 , amer .
soc . , providence , ri , ( 2001 ) 197 - 214 .
lau and s .- m .
ngai , multifractal measures and a weak separation condition , _ adv .
_ , * 141 * ( 1999 ) , 4596 .
n. luzia , a variational principle for the dimension for a class of non - conformal repellers .
@xmath1081 dynam .
_ , * 26 * ( 2006 ) , 821845 .
v. mayer , b. skorulski and m. urbaski , _ distance expanding random mappings , thermodynamical formalism , gibbs measures and fractal geometry _ ,
lecture notes in mathematics , 2036 , springer , heidelberg ( 2011 ) y. pesin and h. weiss , on the dimension of deterministic and random cantor - like sets , symbolic dynamics , and the eckmann - ruelle conjecture , _ comm .
_ , * 182 * ( 1996 ) , 105153 .
y. pesin and h.weiss , a multifractal analysis of equilibrium measures for conformal expanding maps and moran - like geometric constructions , _
j. statist .
_ , * 86 * ( 1997 ) , 233275 .
d. a. rand , the singularity spectrum @xmath1082 for cookie - cutters , _ ergod .
@xmath1081 dynam .
_ , * 9 * ( 1989 ) , 527541 .
o. sarig , thermodynamic formalism for countable markov shifts , _ hyperbolic dynamics , fluctuations and large deviations , proc .
sympos . pure math .
, 89 , amer .
soc . , providence , ri_,(2015 ) , 81117
. l. shu , the multifractal analysis of birkhoff averages for conformal repellers under random perturbations , _ monatsh .
_ , * 159 * ( 2010 ) , 81113 . | we describe the multifractal nature of random weak gibbs measures on some class of attractors associated with @xmath0 random dynamics semi - conjugate to a random subshift of finite type .
this includes the validity of the multifractal formalism , the calculation of hausdorff and packing dimensions of the so - called level sets of divergent points , and a @xmath1-@xmath2 law for the hausdorff and packing measures of the level sets of the local dimension .
zhihui yuan |
historically sasakian structures grew out of research in contact manifolds . in 1960 , sasaki @xcite introduced a geometric structure related to an almost contact structure .
this geometry became known as sasakian geometry and has been studied extensively ever since . in the 1960 s ,
the sasakian structure was studied extensively as an odd dimensional analogue of khler spaces @xcite . afterwards , by developing the khler structure to quaternionic spaces the concept of quaternionic khler and hyperkhler manifolds
were introduced .
hyperkhler manifolds are special class of khler manifolds .
they can be thought of as quaternionic analogues of khler manifolds while a quaternionic khler manifold needs not to be a khler one in general . in recent years
quaternionic khler and hyperkhler manifolds have received a great deal of attention .
they appear and play an important role in many different areas of mathematics and mathematical physics @xcite .
the 3-sasakian manifolds , which have a close relation to both hyperkhler and quaternionic khler manifolds , first appeared in a paper by kuo in @xcite . in 1970 three more papers @xcite were published in the japanese literature discussing sasakian 3-structures . later on , in 1973 ishihara @xcite had shown that if the distribution formed by the three killing vector fields which define the 3-sasakian structure is regular then the space of leaves is a quaternionic khler manifold .
we assume that @xmath3 is a 3-sasakian manifold and @xmath4 denotes the transverse distribution to the riemannian foliation @xmath5 with respect to the metric @xmath6 in the tangent bundle @xmath2 .
the purpose of this paper is to study the geometric properties of the distribution @xmath4 as well as the tangent bundle of a manifold such as curvature tensor , sectional curvature and ricci tensor .
moreover , we investigate the close relation of @xmath1-sectional curvatures of @xmath7 and holomorphic sectional curvatures of @xmath4
. we can refer to @xcite as an analogue work on sasakian manifolds .
aiming at our purpose , we organized this paper as follows . in section 2 , we present some basic notations and definitions which are needed in the following sections . in section 3 , the new linear connection @xmath0 is introduced in the terms of levi - civita connection .
we show that it is a metric connection and completely determined by @xmath4 .
moreover , the @xmath1 structures are parallel with respect to the new connection @xmath0 on @xmath4 . in section 4
, we present the hyperkhler properties of the distribution @xmath4 and call it the hyperkhler contact distribution . in this section , the curvature and ricci tensor of hyperkhler contact distribution @xmath4 is defined with respect to the metric connection @xmath0 . finally , in this section
, we prove some theorems that show the curvature properties of this distribution and its close geometric relation with the 3-sasakian manifold .
let @xmath7 be a @xmath8-dimensional smooth manifold .
then , the triple structure @xmath9 on @xmath7 , consisting of the @xmath10-form @xmath11 , non - vanishing vector field @xmath12 and the 1-form @xmath13 , is called an _ almost contact structure _ if @xmath14 this structure will be called a _ contact structure _ if @xmath15 the manifold @xmath16 with the almost contact or contact structure @xmath9 is called almost contact or contact manifold , respectively , and they are denoted by @xmath17 .
there was proved that every almost contact manifolds admit a compatible riemannian metric in the following sense @xmath18 for all @xmath19 . in case of contact
metric manifolds , the fundamental 2-form @xmath20 defined by @xmath21 coincides with @xmath22 .
let @xmath23 be the levi - civita connection with respect to the metric @xmath6 on the contact manifold @xmath24 .
then , the ( almost ) contact metric manifold @xmath24 is a _
sasakian manifold _
if @xmath25 if @xmath24 is a sasakian manifold then the following equations are satisfied @xmath26 where @xmath27 and @xmath28 are curvature tensor and ricci curvature , respectively , given by following formulas @xmath29}z,\ ] ] @xmath30 where @xmath31 are orthonormal local vector fields on @xmath32 @xcite .
let @xmath32 be a smooth riemannian manifold of dimension @xmath33 .
the manifold @xmath32 is called a 3-_sasakian manifold _ when it is endowed with three sasakian structures @xmath3 for @xmath34 , satisfying the following relations @xmath35 for all even permutation @xmath36 of @xmath37 . for more details about the above definitions see @xcite .
let @xmath38 be the distribution spanned by the three global vector fields @xmath5 . by means of @xmath39 for @xmath34 , one can prove the integrability of @xmath38 as follows @xmath40=\nabla_{\xi_{\alpha}}\xi_{\beta}-\nabla_{\xi_{\beta}}\xi_{\alpha}=2\xi_{\gamma},\ ] ] for all even permutation @xmath41 of @xmath37 .
therefore , @xmath38 defines a 3-dimensional foliation on @xmath7 .
moreover , the equations @xmath42 for @xmath34 show that the foliation @xmath38 is a riemannian foliation .
the transverse distribution of @xmath12 with respect to the metric @xmath6 is denoted by @xmath4 and it is given by @xmath43 .
the distribution @xmath4 is a @xmath44-dimensional distribution on @xmath7 and it decomposes the tangent bundle @xmath2 as follows @xmath45 in a 3-sasakian manifold , the distribution @xmath4 is never integrable and in the sequel we call it 3-_contact distribution_.
consider the foliation @xmath38 and the distribution @xmath4 on a 3-sasakian manifold @xmath7 .
we can choose a local coordinate system by means of foliated charts such that @xmath46 with @xmath47 .
this local frame is adapted to the foliation @xmath38 and one can make the local basis @xmath48 of @xmath4 orthogonal to @xmath38 with respect to the metric @xmath6 from the basis @xmath49 as follows @xmath50 where @xmath51 .
consider local basis @xmath52 of @xmath2 , then the riemannian metric @xmath6 in this local basis will have the presentation as follows @xmath53 where @xmath54 .
now , we evaluate the lie brackets of the local vector fields in ( [ basis1 ] ) .
first , by a direct calculation we obtain @xmath55=\xi_{\alpha}(\eta_i^{\beta})\xi_{\beta}-\eta_i^{\beta}[\xi_{\beta},\xi_{\alpha}],\ \ \ [ \frac{\delta}{\delta x^i},\frac{\delta}{\delta x^j}]=\left(\frac{\delta\eta_i^{\alpha}}{\delta x^j}-\frac{\delta\eta_j^{\alpha}}{\delta x^i}\right)\xi_{\alpha}.\ ] ] then , the lie brackets of @xmath56,@xmath57 and @xmath58 are given by ( [ bracket xi ] ) . by means of @xmath59)=2d\eta^{\beta}(\frac{\delta}{\delta x^i},\xi_{\alpha})=2g(\frac{\delta}{\delta x^i},\varphi_{\beta}(\xi_{\alpha}))=0,\ ] ] and @xmath60),\ ] ] the eq .
( [ bracket1 ] ) will be appeared as follows @xmath61=0,\ \ \ [ \frac{\delta}{\delta x^i},\frac{\delta}{\delta x^j}]=\left(\frac{\partial\eta_i^{\alpha}}{\partial x^i}-\frac{\partial\eta_j^{\alpha}}{\partial x^i}\right)\xi_{\alpha}=-2\omega_{ij}^{\alpha}\xi_{\alpha}.\ ] ] now , we consider tensor fields @xmath62 given by @xmath63 for @xmath64 .
these tensor fields on the 3-sasakian manifold @xmath3 have the following more explicit expressions @xmath65 for all @xmath66 .
[ levi1 ] the levi - civita connection @xmath23 of the riemannian metric @xmath6 on the 3-sasakian manifold @xmath3 has the following components with respect to basis ( [ basis1 ] ) @xmath67 where @xmath68 [ bundle like ] @xcite the foliation @xmath38 is totally geodesic and the riemannian metric @xmath6 is bundle - like for it .
consider the levi - civita connection @xmath23 the on 3-sasakian manifold @xmath32 .
then , we define the linear connection @xmath0 by @xmath69 by direct calculations using ( [ basis1 ] ) and ( [ new conn ] ) we obtain the following theorem . [ new comp . ] locally , the linear connection @xmath0 is completely determined by the formulas @xmath70 from theorem [ new comp . ] and ( [ gamma ] ) , we conclude that @xmath0 is completely determined by riemannian metric induced by the @xmath6 on the 3-contact distribution @xmath4 .
moreover , it is easy to see that the 3-contact distribution @xmath4 and foliation @xmath38 are parallel with respect to the linear connection @xmath0 . for these reasons we call @xmath0 the @xmath4-_connection _ on the 3-sasakian manifold @xmath7 .
+ this is surprising and interesting that the connection @xmath0 presented in ( [ new conn ] ) coincides with the connection @xmath71 defined in @xcite . in the following lemma
, we prove this equality let @xmath32 be a riemannian manifold with a 3-sasakian structure .
then , the connection @xmath0 presented in ( [ new conn ] ) coincides with the connection @xmath71 defined in @xcite . by the definition of @xmath71 in @xcite and using the local frame ( [ basis1 ] ) , we obtain @xmath72 @xmath73=0=\bar{\nabla}_{\xi_{\alpha}}\frac{\delta}{\delta x^i},\ ] ] @xmath74 where @xmath75 denotes the restricted component of the levi - civita connection @xmath23 on 3-contact distribution @xmath4 .
from which , one can find the metrizability of @xmath0 and some more information about its torsion and curvature in @xcite .
however , we present the local expression of its torsion and curvature with respect to the frame ( [ basis1 ] ) as follows @xmath76 for all even permutations @xmath41 of @xmath37 .
@xmath77 where @xmath19 and @xmath78 moreover , the lie brackets of vector fields on @xmath7 in terms of the @xmath4-connection have the following expression @xmath79=\bar{\nabla}_xy-\bar{\nabla}_yx-2\omega^{\alpha}(x , y)\xi_{\alpha},\ ] ] for any @xmath80 . [ varphi ]
consider the linear connection @xmath0 given by ( [ new conn ] ) on 3-sasakian manifold @xmath3 for @xmath34 .
then , the following equation is satisfied @xmath81 where @xmath82 . to complete the proof
, we need to evaluate @xmath83 . using ( [ new conn ] ) , we obtain @xmath84 @xmath85 @xmath86 @xmath87 @xmath88 @xmath89 it is easy to check the last equation is vanish for all @xmath34 .
if we restrict metric @xmath6 and @xmath1 for @xmath34 to the 3-contact distribution @xmath4 , then @xmath4 can be considered as an almost hyper - hermitian vector bundle .
moreover , @xmath1 for @xmath34 are parallel with respect to the metric connection @xmath0 on @xmath4 .
therefore , @xmath4 carries an analogue hyperkhler structure and we call it a _ hyperkhler contact distribution_. the close relation of hyperkhler and 3-sasakian manifolds suggests this name as well . + to study the curvature tensor of 3-contact distribution @xmath4 , we present some properties of this tensor . by corollary
[ bundle like ] , it is deduced that @xmath90 are functions of @xmath91 for @xmath92 ( i.e. @xmath93 for all @xmath34 ) . using this fact and @xmath94=0 $ ] , one can check that @xmath95 and therefore the equations ( [ curvature ] ) and ( [ cur . com . ] ) become @xmath96 where @xmath19 and @xmath97 the equations ( [ curvature1 ] ) and ( [ cur
. com.1 ] ) show that @xmath98 just depends on @xmath4 .
therefore , we put @xmath99 and we call it the _ curvature tensor field _ of @xmath100 .
the first equality in 1 is a general property of curvature tensor of any linear connection .
the next equality of 1 is a consequence of the fact that @xmath0 is a metric connection . to prove 2
, we use the eq .
( [ bracket3 ] ) in a straightforward calculation as follows @xmath103}z\ ] ] @xmath104}x+\bar{\nabla}_z\bar{\nabla}_xy-\bar{\nabla}_x\bar{\nabla}_zy -\bar{\nabla}_{[z , x]}y\ ] ] @xmath105 + 2(\omega^{\alpha}(x,[y , z])+\omega^{\alpha}(\bar{\nabla}_xy , z)+\omega^{\alpha}(y,\bar{\nabla}_xz ) ) \xi_{\alpha}\ ] ] @xmath106 + 2(\omega^{\alpha}(y,[z , x])+\omega^{\alpha}(\bar{\nabla}_yz , x)+\omega^{\alpha}(z,\bar{\nabla}_yx ) ) \xi_{\alpha}\ ] ] @xmath107 + 2(\omega^{\alpha}(z,[x , y])+\omega^{\alpha}(\bar{\nabla}_zx , y)+\omega^{\alpha}(x,\bar{\nabla}_zy ) ) \xi_{\alpha}=0.\ ] ] then , by lemma 3.1 in @xcite on page 32 the item 3 is obtained . finally , taking into account that @xmath1 are parallel with respect to @xmath0 on @xmath4 , the first equality of 4
is obtained .
the next equality is achieved by means of the first one and item 3 .
first we have @xmath108 in term of @xmath109 , for all @xmath110 , as follows @xmath111 and @xmath112}z=\nabla_{[x , y]}z\cr \hspace{1.5cm}+\eta^{\alpha}([x , y])\varphi_{\alpha}(z)+\eta^{\alpha}(z)\varphi_{\alpha}([x , y])+\omega^{\alpha}([x , y],z ) \xi_{\alpha}. \end{array}\ ] ] therefore , @xmath113 where the einstein notation is used for repeated indices @xmath114 and @xmath115 on their range in case of @xmath116 . the _ ricci tensor _ @xmath119 of the hyperkhler contact distribution @xmath4 is defined by @xmath120 where @xmath121 is an orthonormal local basis of vectors in @xmath122 .
then , the local components of ricci tensor @xmath119 are given by @xmath123 in @xcite , it was proved that 3-sasakian manifolds are einstein spaces and their ricci tensor fields with respect to the levi - civita connection are given by @xmath126 where @xmath127 .
+ if @xmath121 is an orthonormal local basis of @xmath4 then @xmath128 will be an unitary orthogonal local basis of @xmath2 , and vice versa . using ( [ sas pro ] ) , ( [ cur1 2 ] ) and ( [ ricci ] ) , for all @xmath82
, we obtain @xmath129 @xmath130 @xmath131 the sectional curvature @xmath132 of levi - civita connection @xmath23 for the plane @xmath133 spanned by @xmath134 at a point is given by @xmath135 it is easy to check that @xmath136 is independent of choosing the vector fields @xmath137 and @xmath138 spanned @xmath133 .
+ the _ holomorphic sectional _ curvatures of the hyperkhler contact distribution @xmath4 are defined by @xmath139 where @xmath117 has unit length at any point with respect to the metric @xmath6 .
[ cons ] let @xmath7 be a 3-sasakian manifold of dimension @xmath33 , with @xmath124 . if only one of the holomorphic sectional curvatures of the hyperkhler contact distribution @xmath4 does not depend on the respective @xmath1-plane @xmath142 , then all holomorphic sectional curvatures are constant on @xmath7 . moreover , this constant value is equal to @xmath143 .
first , without losing the generality , suppose that the holomorphic sectional curvature @xmath144 of @xmath4 is independent of the @xmath145-plane and it is given by @xmath146 where @xmath147 is a function on @xmath7 .
now , let @xmath148 it is easy to check that @xmath149 satisfied ( [ curvature2 ] ) .
moreover , @xmath150 for all unitary vector field @xmath117 .
therefore , we obtain @xmath151 by lemma 4.1 on page 132 of @xcite , the following are obtained @xmath152 now , by calculating the ricci tensor of @xmath149 and using lemma [ ric ] we obtain @xmath153 @xmath154 therefore , @xmath155 from ( [ 1])([3 ] ) , we conclude that @xmath156 where @xmath34 and @xmath137 is an unitary vector field in @xmath4 .
this completes the proof . in theorem
[ cons ] , it is proved that if a holomorphic sectional curvature of @xmath4 is constant then all of them are constant and equal . in this case , we call @xmath4 a hyperkhler contact distribution of _ constant holomorphic sectional _ curvature . by means of this definition and theorem [ cons ]
the following corollary is obtained + [ cons1 ] let @xmath7 be a 3-sasakian manifold of dimension @xmath33 , with @xmath124 .
then the hyperkhler contact distribution @xmath4 is of constant holomorphic sectional curvature @xmath143 if and only if its curvature tensor field @xmath98 on @xmath4 is given by @xmath157 @xmath158 [ cons rela ] let @xmath7 be a 3-sasakian manifold with hyperkhler contact distribution @xmath4
. then @xmath4 is of constant holomorphic sectional curvature @xmath159 if and only if any @xmath1-sectional curvature of @xmath7 for a @xmath1-plane @xmath142 , where @xmath160 , is constant @xmath161 for @xmath34 . up to now
, we use metric connection @xmath0 on hyperkhler contact distribution @xmath4 to define the curvature tensor , holomorphic sectional curvatures and ricci tensor of this distribution .
moreover , in some theorems , we present them in terms of curvature tensor , @xmath1-sectional curvatures and ricci tensor of levi - civita connection on @xmath4 , respectively . in the next theorem ,
we study the sectional curvatures of these two connections on @xmath7 .
[ sec ] let @xmath133 be the @xmath1-plane at a point of a 3-sasakian manifold @xmath7 .
then , the sectional curvatures of @xmath133 with respect to the @xmath23 and @xmath0 are related as follows @xmath162 where @xmath137 is a unit vector in the @xmath1-plane @xmath133 and @xmath163 is an arbitrary permutation of @xmath37 . without losing the generality , we prove these theorem for @xmath164 .
consider the @xmath165-plane @xmath133 and unit vector @xmath140 , then by using ( [ cur1 2 ] ) , we obtain @xmath166 @xmath167 @xmath168 @xmath169 @xmath170 @xmath171 @xmath172 @xmath173 @xmath174 this completes the proof .
1 m. f. atiyah , _ hyper - khler manifolds _ , lwcture notes in math . , 1422 , springer - verlag , berlin - new york , 1990 .
a. bejancu , _ khler contact distribution _ ,
journal geometry and physics , * 60 * ( 2010 ) , 1958 - 1967 .
d. e. blair , _ riemannian geometry of contact and symplectic manifolds _ , birkhauser , boston , 2002 . c. p. boyer , k. galicki _ and _ b. m. mann , _ 3-sasakian manifolds _ , proc .
japan acad . ,
* 69 * ( 1993 ) , 335 - 340 . c. p. boyer , k. galicki _ and
_ b. m. mann , _ the geometry and topology of 3-sasakian manifolds _ , j. reine angew .
, * 455 * ( 1994 ) , 183 - 220 . b. cappelletti montano _ and _ a. de nicola , _ 3-sasakian manifolds , 3-cosymplectic manifolds and darboux theorem _ , * 57 * ( 2007 ) , 2509 - 2520
. n. j. hitchin , _ hyper - khler manifolds , seminaire bourbaki _ ,
* 206 * ( 1992 ) , exp .
748 , 3 , 137 - 166 .
s. ishihara , _ quaternion khlerian manifolds and fiberd riemannian spaces with sasakian 3-structures _ , kodai math .
, * 25 * ( 1973 ) , 321 - 329 .
s. ishihara , _ quaternion khlerian manifolds _ , j. differential geometry , * 9 * ( 1974 ) , 483 - 500 .
t. kashiwada , _ a note on a riemannian space with sasakian 3-structures _ , natural science report , * 22 * ( 1971 ) , 1 - 2
. y. y. kuo , _ on almost contact 3-structure _ , tohoku math .
j. , * 22 * ( 1970 ) , 325 - 332 .
y. y. kuo and s. tachibana , _ on the distribution appeared in contact 3-structure _ , taita j. of math .
, * 2 * ( 1970 ) , 17 - 24 .
m. okumura , _ on infinitesimal conformal and projective transformations of normal contact spaces _
, tohoku math .
j. ( 2 ) , * 14 * ( 1962 ) , 398 - 412 .
s. sasaki , _ on differentieable manifolds with certain structures which are closely related to almost contact structure _ , tohoku math .
j. , * 2 * ( 1960 ) , 459 - 476 .
s. tachibana and w. n. yu , _ on a riemannian space admitting more than one sasakian structure _ , tohoku math . j. , * 22 * ( 1970 ) , 536 - 540 . s. tanno , _ the topology of contact riemannian manifolds _ , illinois j. math . , * 12 * ( 1968 ) , 700 - 717 .
s. tanno , _ on the isometry of sasakian manifolds _ , j. math .
japan , * 22 * ( 1970 ) , 579 - 590 .
k. yano _ and _ m. kon , _ structures on manifolds _ , world scientific publishing , singapore , 1984 . | in this paper , the hyperkhler contact distribution of a 3-sasakian manifold is studied . to analyze the curvature properties of this distribution , the special metric connection @xmath0 is defined .
this metric connection is completely determined by hyperkhler contact distribution .
we prove that hyperkhler contact distribution is of constant holomorphic sectional curvatures if and only if its 3-sasakian manifold is of constant @xmath1-sectional curvatures .
moreover , it is shown that there is an interesting relation between the sectional curvatures of @xmath1-planes on @xmath2 of metric connection @xmath0 and the levi - civita connection . |
many extensions of the standard model predict the existence of light scalar particles . the axion may be the most debated one but there are also approximate nambu - goldstone fields associated with family symmetries , or moduli fields , or dilatons , or superpartners of the gravitino @xcite .
exchange of such particles by ordinary matter will induce forces that extend over the compton wavelength of the particle @xcite .
however , the effect will be felt by bulk matter only if the potential is spin - independent so that forces can add up coherently over macroscopic distances .
now , a pseudoscalar particle , such as the axion , is coupled to fermions via a @xmath2 which , in the nonrelativistic limit , flips the spin .
therefore , single pseudoscalar exchange leads to spin - dependent forces that do not extend over macroscopic unpolarized bodies @xcite . a double exchange of pseudoscalars on the other hand can coherently sum over a macroscopic sample of matter because it can leave the spin unflipped .
the explicit form of these forces has been derived and their phenomenological consequences explored in previous work in the context of nonrelativistic `` old fashioned perturbation theory '' @xcite . here
we reopen the question of pseudoscalar mediated forces in the light of the powerful dispersion theoretical techniques devised by feinberg and sucher and collaborators that make extensive use of full relativistic quantum field theory @xcite .
in section [ sec : metode ] we give the necessary theoretical background which is nothing but a brief resum of the seminal work by feinberg and sucher .
section [ sec : yuk ] is devoted to the yukawa type interaction and section [ sec : der ] deals with derivatively coupled scalars .
we shall see that the two interactions produce quite different potentials @xcite , a fact that could not have been derived in a purely non - relativistic framework @xcite .
the paper ends with a brief summary and conclusions contained in section [ sec : conc ] .
following the general strategy devised by feinberg and sucher @xcite , we define a potential in a given quantum field theory by equating the scattering amplitude for a two body process that follows from the usual feynman rules , with the transition amplitude associated to a schrdinger type equation solved _ la _ lippmann - schwinger .
let us be explicit and consider elastic scattering of particles a and b with four momenta @xmath3 and @xmath4 in the initial state and @xmath5 and @xmath6 in the final state .
the mandelstam variables are then : @xmath7 with @xmath8 . in the c.m .
we write the momenta as @xmath9 now @xmath10 where @xmath11 and @xmath12 with @xmath13 , whose physical region is @xmath14 where @xmath15 \left [ s-{\left ( m_a - m_b \right)}^2 \right ] \right/ 4 s.\ ] ] the transition from initial state @xmath16 to final state @xmath17 is described in quantum field theory by the transition matrix element @xmath18 where @xmath19 are normalization factors of one particle states and @xmath20 is the invariant feynman amplitude .
the definition of our potential follows now from identifying this transition amplitude with @xmath21 where @xmath22 is the sum of the free dirac hamiltonians for particles a and b. here all quantities are referred to the c.m . the feynman amplitude @xmath23 is understood as a series expansion in ( even ) powers of the coupling constant associated to single , double , particle exchange in the t - channel .
we assume that the potential v also admits a series expansion @xmath24 so we determine v order by order in perturbation theory through @xmath25 here we should point out a technicality .
the potential we are after must have the form @xcite @xmath26 where @xmath27 is an operator that projects on the positive energy states of @xmath22 . because @xmath28 and @xmath29 we can rewrite the previous equations that determine the potential as , @xmath30 in principle the above equations permit an iterative determination of the potential to the desired order .
however we are not done yet because we would like to have our potential in position space and what we have is the operator @xmath31 in the momentum representation .
therefore , we should fourier transform our results back to configuration space , i.e. we wish to find @xmath32 such that @xmath33 inversion of the above equation , however , requires knowing the function @xmath23 for all values of three momentum .
but we only know the scattering amplitude on shell , i.e. for @xmath34 .
we can use the fact that @xmath35 is an analytic function of t and so analytically extend its domain beyond the physical region , i.e. for all values of @xmath36 .
suppose @xmath35 is analytic everywhere except for branch cuts on the real axis and furthermore , it vanishes for large @xmath37 .
then , using cauchy s theorem , we can write @xmath38 where @xmath39_t}{2 i}$ ] is the spectral density and @xmath40_t$ ] is the discontinuity of @xmath23 across the cut .
only the piece of the amplitude arising from the right hand cut will be of interest to us for only this piece leads to a long range potential @xcite .
assuming that the basic relations eqs ( [ eq : u2 ] ) and ( [ eq : u4 ] ) hold also in the extended domain , we can fourier invert them as follows . ] , @xmath41 we use now the spectral representation given before to obtain @xmath42 where , to reach this final form , we conveniently changed the order in which integrals were done . in short , obtaining long range potentials amounts to calculating t - channel discontinuities in feynman diagrams and performing a laplace transform .
we shall see how things work out in detail as we do our specific calculations in the next two sections .
our starting point is the lagrangian density , @xmath43 where @xmath44 is a fermion field and @xmath45 is the pseudoscalar field which we take to be massless .
the potential associated to single particle exchange is easily obtained from the discontinuity associated to the diagram in fig .
the spectral density function @xmath46 is in this case @xmath47 after laplace transforming we get the relativistic potential operator @xmath48 where subindices make explicit that dirac matrices act either on spinor a or spinor b. the nonrelativistic limit of equation above leads to the well known spin - dependent potential @xmath49 with @xmath50 and @xmath51 , the masses of particles a and b respectively .
what we are really interested in is @xmath52 , i.e. the potential due to two - particle exchange . to this end
we need the discontinuities of diagrams in fig . 2 and the discontinuity of the subtraction term in eq ( [ eq : u4 ] ) ( iteration of the lowest order potential @xmath53 ) . the feynman amplitude associated to fig .
2 can be written @xmath54 in terms of the compton amplitude , depicted in fig .
3 , @xmath55 u(p ) .\label{eq : comps}\ ] ] making use of the dirac equation and trading the propagators for their discontinuities , i.e. @xmath56 we arrive at @xmath57_t=-\frac{i g^4}{8 \pi^2}\;\int{\!d\phi\;\bar{u}_a ' { } \left [ \frac{p_a\cdot(k'{}-k)\ : \rlap / k}{2\ : p_a \cdot k ' { } \ : p_a\cdot k } \right ] u_a\;\bar{u}_b ' { } \left[\frac{p_b\cdot ( k'{}-k ) \ : \rlap / k}{2\ : p_b \cdot k ' { } \ : p_b\cdot k}\right ] u_b}\ ] ] with the two particle phase space explicitly given by @xmath58 it is convenient to do the integrals in the c.m . of the pseudoscalars ,
i.e. to go to the t - channel , and then use crossing symmetry to recover the original amplitude .
we follow here the notation in reference @xcite where they deal with a related problem .
first define momenta as , @xmath59 introduce next the unit imaginary vectors @xmath60 with @xmath61 and @xmath62 are unitary complex vectors verifying @xmath63 so that all particles are on - shell .
now the discontinuity can be put in the form @xmath57_t=\frac{i
g^4}{4 \pi \ : b t}\;\int{\frac{d\omega}{4 \pi}\ ; \frac{x_a x_b}{d_a d_b}\;\ ; \bar{u}_a ' { } \rlap / k u_a\ :
\bar{u}_b ' { } \rlap / k u_b}\ ] ] where , to simplify expressions , we use @xmath64 the integration to be carried out is an angular average .
we use the shorthand : @xmath65 .
hence , the discontinuity takes the form @xmath57_t=\frac{i g^4}{4 \pi
\:b t}\ ; \bar{u}_a ' { } \gamma_{\mu } u_a\ ; \bar{u}_b ' { } \gamma_{\nu } u_b \:{\cal t}^{\mu \nu}\label{eq : dism4ps}\ ] ] with @xmath66 .
lorentz covariance dictates the following decomposition @xmath67 in terms of the three independent momenta @xmath68 the coefficients @xmath69 can be found to be combinations of scalar integrals as shown in the appendix .
now , in the c.m . of the incident particles , the relations @xmath70 are easily established with the help of the dirac equation .
this leads directly to @xmath71_t = \frac{i g^4}{4 \pi \:b t } \;\bar{u}_a ' { } \bar{u}_b ' { } \left[4 m_a m_b \left ( 2 a_3-a_1-a_2 \right)+4 m_b w \gamma_a^0 \left ( a_2-a_3 \right ) \right .
\nonumber\\ + \left .
\mbox{}4 m_a w \gamma_b^0 \left ( a_1-a_3 \right)+\gamma_a^0 \gamma_b^0 \left ( 4 w^2 a_3+a_4 \right)-{\mbox{\boldmath $ \gamma$}}_a { \mbox{\boldmath $ \gamma$}}_b\ , a_4 \right ] u_a u_b .
\label{eq : disa}\end{aligned}\ ] ] this discontinuity is a complex function since the @xmath69 are complex and hence adds an imaginary component to the spectral density which would finally contribute an imaginary piece to the potential .
inspection of equation ( [ eq : disa ] ) immediately tells us that the offending piece comes from the imaginary parts of the @xmath69 .
but we should recall that we still have to subtract the contribution from the iterated lowest order potential .
it turns out that its imaginary part exactly cancels the unwanted contribution coming from equation ( [ eq : disa ] ) .
indeed we have explicitly checked this to be the case .
however , in order to make this paper not too lengthy , we do not include the intermediate steps of the calculation .
we only report on the result , i.e. @xmath72_t & = & \frac{g^4}{8 p w } \;\bar{u}_a ' { } \bar{u}_b ' { } \left[\frac{t-4 p^2}{(4 p^2+t)^2 } \left ( e_a \gamma_a^0-m_a \right ) \left ( e_b \gamma_b^0-m_b \right)\right .
\nonumber\\ & & - \left.\mbox{}\frac{p^2}{4 p^2+t } { \mbox{\boldmath $ \gamma$}}_a \cdot { \mbox{\boldmath $ \gamma$}}_b \right ] u_a u_b .\label{eq : rm4ps}\end{aligned}\ ] ] the relevant contribution to the long range potential comes from the real parts of the @xmath69 , that is the imaginary part of ( [ eq : disa ] ) , once the contribution of the iterated potential has been subtracted .
let us elaborate on the iteration amplitude , @xmath73 this formal expression can be recast in the explicit form @xmath74 with @xmath75 and @xmath76 and @xmath77 .
the integration over momentum @xmath78 reflects the fact that we have inserted a complete set of plane wave intermediate states in ( [ eq : dmi ] ) .
we have also used @xmath79 where @xmath80 .
the iteration amplitude can be conveniently put as follows @xmath81 where @xmath82 and they can be found in the appendix .
a little bit of dirac algebra and the results in the appendix allow us to write the discontinuity of @xmath83 as @xmath84_t&=&\frac{g^4}{8 \pi^2 } \;\int { \!l^2 d l\ ; \bar{u}_a ' { } \bar{u}_b ' { } \left\ { \left(e_a ' \gamma_a^0-m_a \right ) \left(e_b ' \gamma_b^0-m_b \right ) \right . }
\nonumber\\ & & -\mbox { } 2 \:\frac{p^2+l^2}{4 p^2+t } \left[\left(e_a \gamma_a^0-m_a \right ) \left(e_b ' \gamma_b^0-m_b \right)+\left(e_a ' \gamma_a^0-m_a \right ) \left(e_b \gamma_b^0-m_b \right ) \right]\nonumber\\ & & + \frac{4}{4 p^2+t } \left ( 2 \frac{(p^2+l^2)^2}{4 p^2+t}-l^2 \right ) \left(e_a
\gamma_a^0-m_a \right ) \left(e_b \gamma_b^0-m_b \right)\nonumber\\ & & + \left.{\mbox{\boldmath $ \gamma$}}_a \cdot { \mbox{\boldmath $ \gamma$}}_b \left ( \frac{(p^2+l^2)^2}{4 p^2+t}-l^2 \right ) \right\ } \left[{\cal l}\right]_t u_a u_b \ ; { \cal c } ( p , l ) .
\label{eq : cmi}\end{aligned}\ ] ] the explicit form for @xmath85_{t}$ ] is given in the appendix .
note that in ( [ eq : cmi ] ) we have @xmath86 the dirac delta piece gives a contribution that , as already advertised , will exactly cancel the real part of the fourth order discontinuity function ( [ eq : rm4ps ] ) .
the principal part integral can be cast in the form @xmath87_t=\frac{i g^4}{16 \pi
\sqrt{t } ( 4 p^2+t)^2}\ ; \wp \int_{-1}^1 { \frac{d x}{\sqrt{1-x^2}}\ : { \cal c } ( p , l ) \;\bar{u}_a ' { } \bar{u}_b ' { } \ , { \cal n } } u_a u_b \label{eq : midisn}\ ] ] where we changed the integration variable via the relation @xmath88 \label{eq : defx}\ ] ] and we used the shorthand @xmath89\nonumber\\ & & + \mbox{}4 \left ( 2 ( p^2+l^2)^2-l^2 ( 4 p^2+t ) \right ) \left(e_a \gamma_a^0-m_a \right ) \left(e_b \gamma_b^0-m_b \right)\nonumber\\ & & -\mbox{}{\mbox{\boldmath $ \gamma$}}_a
\cdot { \mbox{\boldmath $ \gamma$}}_b ( 4 p^2+t ) \left ( ( 4 p^2+t ) l^2-(p^2+l^2)^2 \right).\label{eq : n}\end{aligned}\ ] ] it is convenient now to split the function @xmath90 as @xmath91 the integral above can not be done exactly and we will expand the integral in a power series in @xmath92 and @xmath93 .
this is a licit procedure because we will perform a laplace transform that heavily weighs the small t region of the spectral function when determining the long range ( large r ) potential and , we will eventually take the non relativistic limit of the potential , i.e. for @xmath94 .
furthermore , each extra power of @xmath92 or @xmath93 implies a correction to the potential with an extra power of @xmath95 .
we see from equation ( [ eq : sepc ] ) that the calculation will involve doing integrals of the type @xmath96 @xmath97 the explicit results of the integrals needed in our calculation are also given in the appendix .
armed with all this artillery we find for the @xmath98 piece of the discontinuity @xmath99_t$ ] : @xmath100_t&=&\frac{i g^4}{8 \sqrt{t } ( 4 p^2+t)^2 ( m_a+m_b)}\;\bar{u}_a ' { } \bar{u}_b ' { } \nonumber \\ & & \left\ { m_a m_b \left [ 4 t- \frac{2 t p^2 } { m_a m_b}+\frac{p^4 t}{2 m_a m_b } \left ( \frac{1}{m_a^2}+\frac{1}{m_b^2}+\frac{1}{m_a m_b } \right ) \right ] \right.\nonumber\\ & + \mbox{}\gamma_a^0 & m_a m_b \left[-4 t+\frac{2 ( m_a - m_b ) } { m_a^2 m_b } t p^2 \right.\nonumber\\ & & + \left.\mbox { } \frac{t^3}{32 m_a^4}+\frac{t^2 p^2}{4 m_a^4}+\frac{-m_a^3-m_a^2 m_b+m_a m_b^2 + 2 m_b^3}{2 m_a^4 m_b^3 } t p^4 \right ] \nonumber\\ & + \mbox { } \gamma_b^0 & m_a m_b \left[-4 t+\frac{2 ( m_b - m_a ) } { m_a m_b^2 } t p^2 \right.\nonumber\\ & & + \left.\mbox { } \frac{t^3}{32 m_b^4}+\frac{t^2 p^2}{4 m_b^4}+\frac{2 m_a^3+m_a^2 m_b - m_a m_b^2- m_b^3}{2 m_a^3 m_b^4 } t p^4 \right ] \nonumber\\ & + \mbox { } \gamma_a^0 \gamma_b^0 & m_a m_b \left [ 4 t + \left ( \frac{2}{m_a^2}+ \frac{2}{m_b^2}-\frac{2}{m_a m_b } \right ) t p^2\right.\nonumber\\ & & -\mbox{}t^3 \left ( \frac{1}{32 m_a^4}+ \frac{1}{32 m_b^4}+\frac{1}{8 m_a^2 m_b^2 } \right)-t^2 p^2 \left ( \frac{1}{4 m_a^4}+\frac{1}{4 m_b^4}+\frac{1}{m_a^2 m_b^2 } \right ) \nonumber\\ & & -\left.\mbox { } t p^4 \frac{2 m_a^4+m_a^3 m_b+m_a^2 m_b^2+m_a m_b^3 + 2 m_b^4 } { 2 m_a^4 m_b^4 } \right]\nonumber\\ & + \mbox{}{\mbox{\boldmath $ \gamma$}}_a \!\cdot \ ! { \mbox{\boldmath $ \gamma$}}_b&\left .
\left [ 2 t p^2 + \frac{t^2}{2}-\frac{t^2 p^2}{4 m_a m_b}-\frac{t p^4}{m_a m_b } \right ] \right\ } u_a u_b \label{eq : mitc1 } \end{aligned}\ ] ] and for the @xmath101 piece : @xmath102_t&=&\frac{i g^4}{16 \sqrt{t } ( 4 p^2+t)^2 ( m_a+m_b)m_a m_b}\ ; \bar{u}_a ' { } \bar{u}_b ' { } \left\ { \right.\nonumber\\ & \gamma_a^0&m_a m_b \left[\frac{m_a^2+m_a m_b+m_b^2}{16 m_a^3 m_b } t ( 4 p^2+t)^2 \right]\nonumber\\ & + \mbox{}\gamma_b^0&m_a m_b \left[\frac{m_a^2+m_a m_b+m_b^2}{16 m_a m_b^3 } t ( 4 p^2+t)^2 \right]\nonumber\\ & -\mbox{}\gamma_a^0 \gamma_b^0 & \left [ \frac{m_a^4+m_a^3 m_b+2 m_a^2 m_b^2+m_a m_b^3+m_b^4}{16 m_a^2 m_b^2 } t ( 4 p^2+t)^2 \right]\nonumber\\ & -\mbox { } { \mbox{\boldmath $ \gamma$}}_a\ ! \cdot\ ! { \mbox{\boldmath $ \gamma$}}_b & \left .
\left[\frac{m_a^2+m_a m_b+m_b^2}{16 m_a m_b } t ( 4 p^2+t)^2 \right ] \right\ } u_a u_b .
\label{eq : mitc2 } \end{aligned}\ ] ] in both equations above we kept only up to the powers of @xmath92 and @xmath93 that will be needed , either in this section or in the next section , to obtain the leading two - particle exchange potential . in this respect equation ( [ eq : mitc2 ] ) does not contribute to the leading potential just under scrutiny . to obtain the spectral density @xmath103
, we must finally perform the subtraction @xmath104_t-\im
\left[{\cal m}_i\right]_t}{2}. \label{eq : roo}\ ] ] recall that what enters @xmath103 is the real part of the @xmath69 in ( [ eq : disa ] ) .
although the integrals that go in the @xmath69 are exactly given in the appendix , the required subtraction ( [ eq : roo ] ) and final laplace transformation ( [ eq : pott ] ) demand that we here also expand the integrands in a power series in @xmath92 and @xmath93 .
after some lengthy algebra , we arrive at the final form for the imaginary part of @xmath105_t$ ] @xmath106_t = \im \left[{\cal m}\right]_t^{odd } + \im \left[{\cal m}\right]_t^{even}\ ] ] where @xmath107_t^{odd } & = & \frac{g^4 m_a m_b}{4 ( m_a+m_b ) \sqrt{t } ( 4 p^2+t)^2 } \ ; \bar{u}_a ' { } \bar{u}_b ' { } \left\ { \right . \nonumber\\ & & \left [ 2 t-\frac{t p^2}{m_a m_b}+\frac{-3 m_a^4 + 2 m_a^2 m_b^2 - 3 m_b^4}{4m_a^4 m_b^4 } t p^4 \right.\nonumber\\ & & + \mbox{}\frac{-3 m_a^4-m_a^3 m_b+m_a^2 m_b^2-m_a m_b^3 - 3 m_b^4}{8 m_a^4 m_b^4 } t^2 p^2\nonumber\\ & & + \mbox{}\left.\frac{-3 m_a^4-m_a^3 m_b+m_a^2 m_b^2-m_a m_b^3 -3 m_b^4}{64 m_a^4 m_b^4 } t^3 \right]\nonumber\\ & + \mbox { } \gamma_a^0 & \left[-2 t+\frac{m_a - m_b}{m_a^2 m_b } t p^2+\frac{3 m_a+4 m_b}{4 m_a m_b^4 } t p^4 \right.\nonumber\\ & & + \mbox{}\frac{3 m_a^4 + 5 m_a^3 m_b+m_a^2 m_b^2-m_a m_b^3-m_b^4}{8 m_a^4 m_b^4 } t^2 p^2 \nonumber\\ & & + \mbox{}\left.\frac{3 m_a^4 + 5m_a^3 m_b+m_a^2 m_b^2-m_a m_b^3- m_b^4}{64 m_a^4 m_b^4 } t^3 \right]\nonumber \\ & + \mbox { } \gamma_b^0 & \left[-2 t+\frac{m_b - m_a}{m_a m_b^2 } t p^2+\frac{4 m_a+3 m_b}{4 m_a^4 m_b } t p^4 \right.\nonumber\\ & & + \mbox{}\frac{-m_a^4-m_a^3 m_b+m_a^2 m_b^2 + 5 m_a m_b^3 + 3 m_b^4}{8 m_a^4 m_b^4 } t^2 p^2\nonumber\\ & & + \mbox { } \left .
\frac{-m_a^4-m_a^3 m_b+m_a^2 m_b^2 + 5 m_a m_b^3 + 3 m_b^4}{64 m_a^4 m_b^4 } t^3 \right]\nonumber\\ & + \mbox { } \gamma_a^0 \gamma_b^0 & \left [ 2 t+\frac{m_a^2-m_a m_b+m_b^2}{m_a^2 m_b^2 } t p^2 + \frac{t p^4}{m_a^2 m_b^2 } \right.\nonumber\\ & & + \mbox{}\frac{m_a^4+m_a^3 m_b+m_a^2 m_b^2+m_a m_b^3+m_b^4}{8 m_a^4 m_b^4 } t^2 p^2\nonumber\\ & & + \mbox{}\left.\frac{m_a^4+m_a^3 m_b+m_a^2 m_b^2+m_a m_b^3+m_b^4}{64 m_a^4 m_b^4 } t^3 \right]\nonumber\\ & + \mbox { } { \mbox{\boldmath $ \gamma$}}_a \cdot { \mbox{\boldmath $ \gamma$}}_b & \left[\frac{p^2
t}{m_a m_b}+\frac{t^2}{4 m_a m_b}+\frac{m_a^2 + 4 m_a m_b+m_b^2}{2 m_a^3 m_b^3 } t p^4\right.\nonumber\\ & & + \left.\mbox{}\frac{3 m_a^2 + 5 m_a m_b+3 m_b^2}{8 m_a^3 m_b^3 } t^2 p^2+\frac{2 m_a^2 + m_a m_b+2 m_b^2}{32 m_a^3 m_b^3 } t^3 \right]\nonumber\\ & \left .
\phantom{\left[\frac{p^2}{m_a}\right ] } \right\ } & u_a u_b \label{eq : carroim}\end{aligned}\ ] ] contains the odd powers of @xmath108 and @xmath109_t^{even } & = & \frac{g^4}{4 \pi}\ ; \bar{u}_a ' { } \bar{u}_b ' { } \left\ { -\frac { ( m_a + m_b ) \gamma_b^0 -m_b}{6 m_a^2 m_b } -\frac { ( m_a + m_b ) \gamma_a^0 -m_a}{6 m_a m_b^2}+\frac{\gamma_a^{\mu } \gamma^b_{\mu}}{4 m_a m_b } \right.\nonumber\\ & & -\left.\mbox{}\frac{m_a m_b+\left ( ( m_a+ m_b ) \gamma_a^0-m_a \right ) \left ( ( m_a + m_b ) \gamma_b^0-m_b \right)}{12 m_a^2 m_b^2 } \right\ } u_a u_b \label{eq : mres}\end{aligned}\ ] ] contains the even powers of @xmath108 .
we did this separation to emphasize that , after the subtraction in equation ( [ eq : roo ] ) , only the term ( [ eq : mres ] ) survives to leading non - vanishing order .
indeed , equation ( [ eq : carroim ] ) coincides exactly with equation ( [ eq : mitc1 ] ) if we neglect terms beyond @xmath110 , or @xmath111 .
the final step involves the laplace transformation indicated by equation ( [ eq : pott ] ) . using the general formula @xmath112 we get @xmath113 which leads , in the non relativistic limit and concentrating only on the spin - independent terms of ( [ eq : u4psrel ] ) , to the long - range attractive potential : @xmath114 where this operator is supposed to act between two - component pauli spinors .
in this section we consider the interaction lagrangian @xmath115 which is how goldstone bosons couple to fermions .
this derivative coupling leads to the same one particle exchange feynman amplitude as before and therefore to the same lowest order potential ( [ eq : u2ps ] ) .
hence the iteration amplitude will be also identical .
however , the two particle exchange amplitude ( see , fig .
2 ) is different because the compton amplitude that goes into ( [ eq : m4c ] ) is different .
indeed , the compton amplitude , corresponding to fig .
3 , is now , @xmath116 u(p).\ ] ] this amplitude differs from ( [ eq : comps ] ) by an extra term proportional to @xmath117 .
we introduce this amplitude in ( [ eq : m4c ] ) and replace the massless propagators by dirac deltas to obtain the discontinuity function : @xmath71_t=&-&\frac{i g^4}{8 \pi^2}\;\int{\!d\phi\;\bar{u}_a ' { } \left [ \frac{p_a\cdot(k'-k)\rlap / { k}}{2\ : p_a \cdot k ' \ : p_a\cdot k } \right ] u_a\;\bar{u}_b ' { } \left[\frac{p_b\cdot ( k'-k ) \rlap / { k}}{2\ : p_b \cdot k ' \ : p_b\cdot k}\right ] u_b}\nonumber\\ & -&\frac{i g^4}{8 \pi^2}\;\int{\!d\phi\,\left\ { \bar{u}_a ' { } u_a\ : \bar{u}_b ' { } u_b \ ; \frac{1}{m_a m_b}-\frac{1}{m_a}\ ; \bar{u}_a ' { } u_a\ : \bar{u}_b ' { } \left [ \frac{p_b \cdot ( k'- k)}{2 \ : p_b \cdot k \ : p_b \cdot k ' } \rlap / { k } \right ] u_b\right.}\nonumber\\ & & -\mbox { } \left .
\frac{1}{m_b}\ ; \bar{u}_a ' { } \left [ \frac{p_a \cdot ( k ' - k ) } { 2\ : p_a \cdot k \ : p_a \cdot k ' } \rlap / { k } \right ] u_a
\bar{u}_b ' { } u_b \right\}.%\label{eq : delm4 } \end{aligned}\ ] ] the first piece is exactly what we had in the last section .
we call @xmath118_{t}$ ] the extra added piece that involves the integrations @xmath119 and @xmath120 } u_b \quad + \quad \left ( a \leftrightarrow b \right)\nonumber\\ & & = \frac{i g^4}{8 \pi^2 m_a}\ ; \int{\ ! d \phi \ ; \bar{u}_a ' { } u_a \ : \bar{u}_b ' { } \rlap / { k } u_b \ , \frac{i \sqrt{t } \xi_b m_b x_b}{\frac{t}{2 } \left(\frac{t}{4}+(\xi_b m_b x_b)^2 \right)}}\quad + \quad \left ( a \leftrightarrow b \right)\nonumber \\ & & = - \frac{g^4}{8 \pi m_a \sqrt{t } \xi_b m_b } \ ; \bar{u}_a ' { } u_a \ : \bar{u}_b ' { } \gamma_{\mu } u_b \ , \left\langle \frac{x_b}{d_b}\ : k^{\mu } \right\rangle \quad + \quad \left ( a \leftrightarrow b \right ) \nonumber\\ & & = \frac{i g^4}{16 \pi m_a m_b \xi_b^2 } \left ( 1-\tau_b \arctan \left(\frac{1}{\tau_b } \right ) \right ) \ : \bar{u}_a ' { } u_a \:\bar{u}_b ' { } u_b \quad + \quad \left ( a \leftrightarrow b \right ) .
\label{eq : dels2 } \end{aligned}\ ] ] the last line in equation ( [ eq : dels2 ] ) is reached by demanding lorentz covariance to write @xmath121 by solving for the coefficients as explained in the appendix , and by using the dirac equation . putting things together , @xmath122_t&=&\frac{i g^4}{16
\pi m_a m_b } \ ; \bar{u}_a ' { } u_a \ : \bar{u}_b ' { } \left[\frac{1}{\xi_a^2 } \left(1-\tau_a \arctan \left(\frac{1}{\tau_a } \right ) \right)\right.\nonumber\\ & & + \mbox{}\left.\frac{1}{\xi_b^2 } \left(1-\tau_b \arctan \left(\frac{1}{\tau_b } \right ) \right)-1 \right ] u_a u_b \label{eq : dm4}\end{aligned}\ ] ] to leading order in @xmath92 and @xmath93
we have @xmath123_t=\frac{i g^4}{16 \pi m_a m_b } \ ; \bar{u}_a ' { } u_a \ : \bar{u}_b ' { } u_b .
% \label{eq : afm4}\ ] ] in the nonrelativistic limit this contributes the quantity @xmath124 which exactly cancels contribution ( [ eq : v4psnr ] ) , i.e. @xmath125 hence , to find the form for the potential in the case under scrutiny , we must go to the next order in our series expansions .
what we need now is to consistently take into account the previously neglected terms in the spectral density @xmath126_t}{2i}.\ ] ] so we collect the relevant pieces in ( [ eq : carroim ] ) , ( [ eq : mitc1 ] ) , and ( [ eq : mitc2 ] ) in addition to @xmath127_t^{(4 ) } \sim \ldots - \frac{i g^4 \sqrt{t}}{16 m_a m_b}\ ; \bar{u}_a ' { } \bar{u}_b ' { } \left ( \frac{1}{4 m_a}+\frac{1}{4 m_b } \right ) u_a u_b % \label{eq : delr4}\ ] ] which is the next to leading term in the expansion of ( [ eq : dm4 ] ) . the result is @xmath128_t } { 2 i}\nonumber\\ & = & \frac{g^4 m_a m_b \sqrt{t}}{8 ( m_a+m_b)}\ ; \bar{u}_a ' { } \bar{u}_b ' { } \ , \left\{\right.\nonumber \\ & & -\mbox{}\frac{3 m_a^4 + 5 m_a^3 m_b+7 m_a^2 m_b^2 + 5 m_a m_b^3 + 3 m_b^4}{64 m_a^4 m_b^4}\nonumber\\ & & + \mbox{}\gamma_a^0 \frac{3m_a^4 + 5 m_a^3 m_b-2 m_a m_b^3 - 3 m_b^4}{64 m_a^4 m_b^4}\nonumber\\ & & + \mbox{}\gamma_b^0 \frac{-3m_a^4 - 2 m_a^3 m_b+5 m_a m_b^3 + 3 m_b^4}{64 m_a^4 m_b^4}\nonumber\\ & & + \mbox{}\gamma_a^0 \ , \gamma_b^0 \ , \frac{3m_a^4 + 2 m_a^3 m_b+7 m_a^2 m_b^2 + 2 m_a m_b^3 + 3 m_b^4}{64 m_a^4 m_b^4}\nonumber\\ & & \left.+ \mbox{}\mbox{spin - dependent terms of the kind}\ ; { \mbox{\boldmath $ \gamma$}}_a \cdot { \mbox{\boldmath $ \gamma$}}_b \right\}\ , u_a u_b .
\label{eq : ror4}\end{aligned}\ ] ] this spectral density nonetheless gives a vanishing spin - independent potential in the static approximation , i.e. @xmath129 the first non - vanishing contribution to the spin - independent potential arises from the part in the spectral density which is linear in @xmath92 .
indeed , the explicit form of the spectral density reads : @xmath130_t}{2 i}=\frac{g^4}{8 \pi } \ ; \frac{1}{240 m_a^4 m_b^4 } \;\bar{u}_a ' { } \bar{u}_b'{}\,\left\ { \right.\nonumber\\ & & 4\left [ \left ( -3 m_a^4 - 2 m_a^3 m_b+2 m_a^2 m_b^2 - 2 m_a m_b^3 - 3 m_b^4 \right ) p^2 \right.\nonumber\\ & & \left.+\mbox { } \left(6 m_a^4 + 3 m_a^3 m_b+4 m_a^2 m_b^2 + 3 m_a m_b^3 + 6 m_b^4 \right ) t \right]\nonumber \\ & + \gamma_a^0 & 2 ( m_a+m_b ) \left [ \left ( 6 m_a^3 - 2 m_a^2 m_b+3 m_a m_b^2 - 4 m_b^3 \right ) p^2 \right.\nonumber\\ & & \left.+\mbox { } \left(-12 m_a^3 + 6 m_a^2 m_b-4 m_a m_b^2 + 3 m_b^3 \right ) t \right]\nonumber\\ & + \gamma_b^0 & 2 ( m_a+m_b ) \left [ \left ( -4 m_a^3 + 3 m_a^2 m_b-2 m_a m_b^2 + 6 m_b^3 \right ) p^2 \right.\nonumber\\ & & \left.+\mbox { }
\left(3 m_a^3 - 4 m_a^2 m_b+6 m_a m_b^2 - 12 m_b^3 \right ) t \right]\nonumber\\ & + \gamma_a^0 \gamma_b^0 & \left [ \left ( 8 m_a^4 + 2 m_a^3 m_b-12 m_a^2 m_b^2 + 2 m_a m_b^3 + 8 m_b^4 \right ) p^2 \right.\nonumber\\ & & \left . \left.+\mbox { } \left(-6 m_a^4 + 2 m_a^3 m_b-9 m_a^2 m_b^2 + 2 m_a m_b^3 - 6 m_b^4 \right ) t \right ] \right\}\ , u_a u_b \label{eq : ror5}\end{aligned}\ ] ] where no iterated second order amplitude contributes to this order , and where we picked the term proportional to @xmath92 in the expansion of ( [ eq : dels1 ] ) and ( [ eq : dels2 ] ) .
if we use now @xmath131 and pass to the static limit , we find @xmath132 which , upon laplace transformation , leads to @xmath133 for the desired spin - independent long range potential .
note that , as opposed to the yukawa type coupling potential ( [ eq : v4psnr ] ) , the derivative interaction leads to a repulsive potential .
very light particles can mediate forces extending over distances on the order of their compton wavelength . if this range is macroscopic , unpolarized bulk matter will only experience the effect of spin - independent interactions .
it is a well known fact that the yukawa potential due to pseudoscalar exchange depends on spin and as a consequence no coherent effects do arise on a macroscopic scale , unless of course our sample is polarized @xcite .
however , residual ( van der waals type ) forces may arise between macroscopic bodies in the case of pseudoscalar mediated interactions , due to the exchange of two quanta `` at the same time '' .
the double helicity flip involved eventually makes the resulting effective potential spin - independent @xcite . in the preceding sections we have established , with the help of the formalism developed by feinberg and sucher , the large distance behavior of such residual forces , i.e. those associated to double pseudoscalar exchange
we have considered two different basic couplings of those scalars to matter fermions . on the one hand we took the ordinary yukawa coupling ( e.g. ,
this is the way the higgs particle couples to fermions ) and on the other we considered the derivative coupling ( e.g. , the axion - fermion interaction ) .
both couplings reduce to the same spin - flip interaction in the static non - relativistic limit . and both interactions also produce identical one - particle exchange ( spin - dependent ) potentials . in spite of this fact , we have explicitly shown that the spin - independent two - particle exchange potential is substantially different in both cases .
indeed , for yukawa coupling we derive a @xmath0 attractive long distance behavior whereas for the derivative coupling the potential , that is now repulsive , falls off as @xmath1 .
since , as emphasized , double exchange will lead to residual macroscopic effects , these effects will be quite different in both cases .
so , we have found still another instance where the interaction of pseudoscalars to fermions can be discriminated .
other places are , soft pion emission in proton - proton scattering or axion bremsstrahlung in a supernova core @xcite . of course , the effects just reported are extremely small for the light scalars presently contemplated in particle physics , such as the axion , and thus their experimental detection is beyond reach of present technology . however , there is much activity and interest on the experimental front and experiments are designed and performed that explore the sub - centimeter and sub - millimeter regime with an ever increasing sensitivity @xcite . and , on the theoretical side , the completion of the particle physics paradigm may still require new superlight scalar particles to exist .
work partially supported by the cicyt research project aen950882 .
f.f . acknowledges the cirit for financial support .
the angular integrals @xmath137 are given next .
the explicit calculation of @xmath137 for @xmath138 is given in @xcite @xmath139 can be found in @xcite and @xmath140 follows trivially from @xmath139 by interchanging @xmath141 and @xmath142 .
for the first two integrals from the set ( [ eq : defij ] ) one gets @xmath143 where we have defined , @xmath144 and the rest is given in terms of @xmath145 and @xmath146 by @xmath147 we start with ( [ eq : dl ] ) .
its discontinuity @xmath148_t$ ] has been obtained in @xcite : @xmath149_t = i \left [ \frac{\pi}{l \sqrt{t } } \frac{1}{\sqrt { \left(l_+^2-l^2 \right ) \left(l^2-l_-^2 \right ) } } \right ] \theta \left(l_+^2-l^2 \right ) \theta \left(l^2-l_-^2 \right)\label{eq : disl}\ ] ] with @xmath150 for @xmath151 , simplifies to @xmath152_t \right|_{p^2=l^2}=\frac{i \pi}{p^2\ : t}.\label{eq : dislon}\ ] ] finally let us discuss ( [ eq : dup ] ) .
this integral admits the general decomposition @xmath159 in terms of the vectors @xmath160 , defined above , and @xmath161 .
clearly , we can write @xmath162 ( we used : @xmath163 ) . the scalar integrals on the left are , respectively @xmath164\nonumber\\ & & + \mbox{}\frac{4 p^2+t}{4 p^2}+\frac{p^2+l^2}{16 p^3
l } t \ln \left[\frac{(p+l)^2}{(p - l)^2 } \right ] \nonumber\\ q_i q_j \upsilon^{i j } & = & \frac{1}{4 } \left ( -2+\frac{2 p^2+t}{p^2}+\frac{p^2+l^2}{4 p^3
l } t \ln \left[\frac{(p+l)^2}{(p - l)^2 } \right ] \right)\nonumber\\ \delta_{ij}\upsilon^{i j } & = & l^2 { \cal l}.\end{aligned}\ ] ] now the three equations can be solved for @xmath141 , @xmath142 , and @xmath165 . recall that we only need the discontinuity of ( [ eq : dup ] ) , i.e. the discontinuities of @xmath141 , @xmath142 , and @xmath165 .
they are , @xmath166_t & = & \frac{1}{4 p^2+t } \left ( 2 \frac{(p^2+l^2)^2}{4 p^2+t}-l^2 \right ) \left[{\cal
l } \right]_t\nonumber\\ \left [ b \right]_t & = & \frac{\left [ c\right]_t}{t}\ ; \left[{\cal l } \right]_t\nonumber\\ \left [ b \right]_t & = & \left(l^2-\frac{(p^2+l^2)^2}{4 p^2+t } \right)\left[{\cal l } \right]_t .
\label{eq : disup } \end{aligned}\ ] ] | using dispersion theoretic techniques , we consider coherent long range forces arising from double pseudoscalar exchange among fermions . we find that yukawa type coupling leads to @xmath0 spin independent attractive potentials whereas derivative coupling renders @xmath1 spin independent repulsive potentials . |
polarimetric wide - field imaging is a recent and important trend in radio astronomy .
many new and planned radio telescopes such as lofar , lwa , mwa , and ska have a wide field - of - view ( fov ) as a major feature .
technologically , this has been made possible by the development of phased dipole arrays and focal plane arrays , both of which have an inherently wider fov than traditional single - pixel radio telescopes .
the motivation for wide fov polarimetric radio telescopes in astronomy is that they facilitate the study of polarized phenomena not restricted to narrow fields such as the large , highly - structured , polarized features discovered in recent polarimetric galactic surveys @xcite . with lofar now producing its first _ all - sky _ images @xcite , a new era of polarimetric wide - field imaging is starting in radio astronomy . despite this trend ,
a complete theory for wide - field , polarimetric astronomical interferometry is lacking . ultimately , classical interferometry is based on the far - zone form of the van cittert - zernike ( vc - z ) theorem @xcite which
, in its original form , is a scalar theory and only valid for narrow fields .
such restrictions are obviously unacceptable in astronomy where polarization often provides crucial astronomical information , and sources are distributed on the celestial sphere , not necessarily limited to small patches . in radio astronomy , a polarimetric , or vector ( i.e. non - scalar ) , extension of the narrow - field vc - z theorem
was given by @xcite , and more recently this was given a consistent mathematical foundation by @xcite in the form of the so - called _ measurement equation _
( m.e . ) of radio astronomy
. on the other hand , a wide - field extension of the scalar vc - z theorem for radio astronomy was given by @xcite , and more recently , imaging techniques for scalar , wide - fields have been an active field of research @xcite .
however , a generalised vc - z theorem for radio astronomy that is both polarimetric and wide - field has not been derived
. this may be because it has been assumed that such a theorem would be a trivial vector- or matrix - valued analogue of the wide - field , scalar theory @xcite . as we will see , however , the final result is not that simple .
the purpose of the present work is therefore to derive a vc - z type relation by generalising the standard m.e .
formalism to allow for arbitrarily wide fields .
this is achieved by generalising the two - component jones formalism to a three - component wolf formalism @xcite . in
what follows , we will derive a vc - z relation that is valid for the entire celestial sphere and is fully polarimetric .
we will show that the standard m.e . can be recovered through a two - dimensional projection .
we will also show that a dual - polarized interferometer of short electric dipoles ( hertzian dipoles ) is inherently aberrated polarimetrically .
we also extend our vc - z relation to include not only the full electric field , but also the full magnetic field , and thereby establish the complete second - order statistical description of the electromagnetic relation between source terms and interferometer response .
our wide - field vc - z relations should be of relevance to the subject of _ direction - dependent effects _ in radio interferometric imaging , which has attracted recent attention @xcite .
the vc - z theorem as used in astronomy is different from its original use in optical coherence . in astronomy one images sources on the celestial sphere based on localised measurements of their far - fields
this is possible because the vc - z theorem provides an explicit relationship between the visibility measured directly by an interferometer and the brightness distribution of the sources ( * ? ? ?
* chap 14 ) .
this makes the vc - z relationship the foundation of synthesis mapping and interferometric imaging . despite its importance
, the vc - z relations in use in astronomy are all either based on the paraxial approximation ( narrow fov ) or they take the source emissions to be scalar .
these simplifications are questionable when the sources cover wide fields or are highly polarized , as is often the case in radio astronomy . here
we derive the full vector electromagnetic analogue to the wide - field , scalar vc - z relation derived in ( * ? ? ? * chap 14 ) , and , as we will see , the final result is not simply a matrix- or vector - valued version of the scalar relation .
we seek a relationship between the electromagnetic field coherence of a distribution of radio astronomical sources and the resulting electromagnetic field coherence at a radio astronomical interferometer . to simplify the discussion
, we will use the term _
interferometer _ as a shorthand for radio astronomical interferometer .
the treatment is intended for earth - based interferometer observations , but it also has space - based radio astronomical interferometry in mind . consider the problem illustrated in fig .
[ fig : coordsys]a ) .
radio emissions from far away sources , such as the point source @xmath0 , are measured by an interferometer located in the domain @xmath1 .
we want to establish a relationship between the coherence of the electric field emanating from a source distribution and the coherence of the electric field in @xmath1 .
let us first consider a single point source @xmath0 located at @xmath2 .
the source is thus at a distance @xmath3 in the direction given by the unit vector ( direction cosines ) @xmath4 with respect to the origin @xmath5 approximately at the centre of @xmath1 .
the source is within the interferometers fov @xmath6 which is centred on point @xmath7 on the celestial sphere .
the electric field from @xmath0 is measured by the interferometer at pairs of positions @xmath8 and @xmath9 in @xmath1 .
@xmath1 is assumed to be bounded , so the maximum distance between any two measurement points @xmath10 ( maximum baseline ) is finite .
although the source emission may be broadband we split it into narrow , quasi - monochromatic , spectral bands and consider a typical narrow ( bandwidth much smaller than centre frequency ) band centred on frequency @xmath11 .
the assumption that @xmath0 is very far away , which quantitatively we take to mean that @xmath12 for all @xmath13 , implies that the entire interferometer is in the far - field of the source .
this means that the electric field @xmath14 at point @xmath8 at time @xmath15 is @xmath16 where @xmath17 and @xmath18 is the complex electric field amplitude vector emitted by the source at @xmath19 in the direction of @xmath8 . furthermore , in the far - field , @xmath20 is approximately transverse so that @xmath21 if we further assume that the angular extent of @xmath1 as seen from @xmath0 is small , that is @xmath22 then the interferometer is in the fraunhofer far - field of the sources . in this case
we can use the approximations @xmath23 and @xmath24 to simplify the expression for the electric field in @xmath1 to @xmath25 where @xmath26 here we have dropped the first argument of @xmath20 since we are assuming that the angular variation of the sources emissions is small enough so that the @xmath23 approximation implies that @xmath27 . a similar expression to equation ( [ eq : diselcontribvecapprox ] ) for the field at @xmath9
is obtained by replacing @xmath8 with @xmath9 and @xmath28 with @xmath29 .
the electric coherence matrix ( tensor ) @xmath30 can then be found by taking the outer product of the electric fields @xmath31 where @xmath32 denotes complex conjugation , @xmath33 denotes time averaging , and the subscripts @xmath34 label cartesian components @xmath35 , which we will define more precisely later .
we have written the electric coherence matrix , equation ( [ eq : ecohmat ] ) , as a @xmath36 complex matrix even though , for the single point source we are considering here , its rank is two and could therefore be expressed as a @xmath37 complex matrix .
we keep the electric coherence matrix as a @xmath36 complex matrix since it is valid even when there are more than one point source .
we now move to the case of a finite number of point sources .
the total electric field measured at points @xmath8 and @xmath9 is now the sum of the fields from sources in directions @xmath38 for @xmath39 .
the electric coherence matrix is therefore @xmath40 for @xmath41 . in going from the double to the single sum we used the usual vc - z assumption that the sources are spatially incoherent , that is , sources in different directions are statistically independent . until now we have considered only discrete sources .
we can make the transition to the more general continuum source distribution by introducing the three - dimensional brightness matrix @xmath42 as a function of direction @xmath19 in a continuous source distribution @xmath43 where @xmath44 is the infinitesimal area of the source distribution .
the superscript is to highlight the fact that @xmath42 is three - dimensional as opposed the usual two - dimensional brightness matrix @xcite .
the reason that the brightness matrix here is three - dimensional is simply because the full electric field amplitude @xmath20 is three - dimensional .
however , due to equation ( [ eq : transcondapprox ] ) , not all the components of @xmath42 are arbitrary . in fact , we will show that it can be recast as one two - dimensional matrix . by making the replacement ( [ eq : defbrithree ] ) in equation ( [ eq : discretesrcecorr ] )
, we obtain @xmath45 where @xmath46 is an infinitesimal solid angle of the source distribution in @xmath6 . in this expression
we see that the dependence on the pairs of position is only relative , that is , the coherence matrix depends only on @xmath47 , and so we recast the expression in terms of the vector @xmath48 also known as the baseline vector measured in wavelengths ( @xmath49 ) .
in practice , rather than use @xmath30 directly , it is convenient and conventional to put the phase reference point at the centre of the fov given by the direction @xmath50 .
the result of the change in phase , @xmath51 is the 3@xmath523 generalisation of the standard 2@xmath522 visibility matrix as defined by , for instance , @xcite .
@xmath53 is a general complex matrix except for @xmath54 where it is hermitian .
it fulfill the symmetry relation @xmath55 where @xmath56 stands for hermitian transpose .
if we use @xmath53 in equation ( [ eq : continumize ] ) we arrive at @xmath57 equation ( [ eq : vcz3dimpl ] ) is a matrix version of the wide - field , scalar vc - z relation and so includes the partial polarization of the ( non - scalar ) source distribution . however , it is not very useful in this form since it does not automatically fulfill the constraint ( [ eq : transcondapprox ] ) .
when applied to @xmath42 , this constraint becomes @xmath58 for all directions @xmath19 , where @xmath19 is understood to be a column vector .
the constraint can , however , easily be removed if we express @xmath20 in terms of spherical base vectors and set the radial component to zero .
this leads us to introduce two coordinate systems : a cartesian and a spherical .
we use the angular , or tangential , basis set @xmath59 of a spherical polar coordinate system as the basis for the polarization of the transverse field of the source distribution on the celestial sphere , see fig .
1b ) . to simplify the results , the spherical coordinate system is taken relative to the phase reference position of the interferometer assumed to be at @xmath7 . in other words , the zero point of the spherical system , @xmath60 , ( intersection of the equator and central meridian ) , is taken to coincide with @xmath7
@xmath61 is the angle from the equator ( positive in the hemisphere with the pole @xmath62 and negative in the other hemisphere ) , and @xmath63 is the position angle from the central meridian around @xmath64 in the anticlockwise sense looking along @xmath64 . in lieu of any other reference directions ,
the orientation of the spherical system around @xmath7 is arbitrary , but for earth - based measurements @xmath62 could be directed towards the north pole or zenith .
one should note that @xmath59 are consistent with ludwig s second definition as detailed in @xcite with the understanding that the antenna boresight in @xcite is here at @xmath7 , that ludwig s reference polarization unit vector @xmath65 is here @xmath66 , and ludwig s cross polarization unit vector @xmath67 is here @xmath68 .
see @xcite for the use of ludwig s third definition in a vc - z relation .
the cartesian system , with base vectors @xmath69 , is defined so that @xmath70 is in the direction of @xmath7 , and @xmath71 is in the direction of the pole @xmath62 . in terms of the cartesian system , we can explicitly write the components of the vectors in equation ( [ eq : vcz3dimpl ] ) as @xmath72 where @xmath73 and where the superscript @xmath74 stands for vector transpose .
all these vectors are unit vectors with real - valued components ranging between @xmath75 and @xmath76 .
these definitions of the @xmath77 and @xmath78 spaces are the same as the usual definitions for earth - based observations , see e.g. , @xcite .
also the matrices in equation ( [ eq : vcz3dimpl ] ) are to be considered in what follows as being expressed in the cartesian system .
the relationships between the spherical and cartesian systems base vectors are @xmath79 using these spherical and cartesian systems we can express the three - dimensional transverse electric field as @xmath80 where @xmath81 is the jones vector in spherical ( rather than the usual cartesian ) components and @xmath82 is the @xmath83 transformation matrix between the components given by the equations ( [ eq : phixyz ] ) and ( [ eq : thetaxyz ] ) .
note that this transformation is possible for all directions on the celestial sphere except for @xmath84 , i.e. , the poles of the spherical coordinate system .
if one wishes to use to a polar spherical system in which the centre of the fov is not on the equator , as it is here , but rather at some declination @xmath85 , then one simply replaces @xmath86 with @xmath87 defined in appendix [ sec : coordtrans ] , equation ( [ eq : sph2cart_gen ] ) .
this assumes that the pole @xmath62 is towards the earth s north pole .
it is easy to show that @xmath88 so the transverse electric field expressed according to equation ( [ eq : cartastrans ] ) does indeed fulfill equation ( [ eq : transcondapprox ] ) .
so if we use @xmath89 rather than @xmath20 in equation ( [ eq : vcz3dimpl ] ) we would have an unconstrained vc - z equation .
we can introduce this replacement by rewriting @xmath42 using equation ( [ eq : cartastrans ] ) , so @xmath90 where , suppressing the dependence on @xmath91 , @xmath92 is the 2@xmath522 brightness matrix , but in spherical rather than cartesian coordinates . by this we mean that , for an arbitrary direction @xmath91 , @xmath93 is locally equivalent to the usual paraxial brightness matrix in cartesian coordinates . from its definition it easy to see that @xmath93 is a hermitian matrix . by using the @xmath78 and @xmath77 spaces , as spanned by the vectors @xmath19 , and @xmath94
, we can write equation ( [ eq : vcz3dimpl ] ) in a more explicit form . the exact form , though , depends on the extent of @xmath6 .
if it is entirely in the hemisphere @xmath95 , then we write @xmath96 and @xmath97}}{n } \,\mathrm{d}l\mathrm{d}m , \label{eq : vcz3d}\ ] ] where we have used @xmath98 , and where the matrices depend implicitly on @xmath99 and @xmath100 .
if , however , part of @xmath6 is in the @xmath101 hemisphere , then we must add to equation ( [ eq : vcz3d ] ) the contribution from this hemisphere given by the integral @xmath102}}{|n| } \,\mathrm{d}l\mathrm{d}m , \label{eq : vcz3dmin}\ ] ] where @xmath103 is the subset of @xmath6 in the @xmath101 hemisphere .
the image horizon , @xmath104 , can also be included by reparametrising the integral in terms of @xmath105 rather than @xmath91 and using the replacement @xmath106 .
the poles @xmath107 can also be imaged , but one must then stipulate the orientation of the @xmath68 and @xmath66 vectors at these singular points .
now , by extending @xmath108 to cover the entire @xmath95 hemisphere and extending @xmath108 to cover the entire @xmath101 hemisphere , the entire celestial sphere can be imaged in a single telescope pointing .
an assumption here is that @xmath1 is a proper three - dimensional volume , or in other words , the baselines should be non - coplanar . if @xmath1 is just a plane ( coplanar baselines )
then only one hemisphere can be mapped uniquely .
equation ( [ eq : vcz3d ] ) is our main result .
it says that the full @xmath36 electric visibility matrix @xmath53 on the three - dimensional @xmath77 space is given by the @xmath37 brightness matrix @xmath93 on the two - dimensional @xmath109 plane .
the fact that this is a relationship between two matrices with different matrix dimensions is a fundamental feature of our vc - z , and makes it clear that it is not just a matrix - valued generalisation of the wide - field , scalar vc - z .
mathematically , this is ultimately due to the ranks of the fundamental matrices , of which we will speak more in section [ sec : stokesparams ] .
the remaining vc - z relationships that form a complete characterisation of the electromagnetic coherence response of a radio astronomical interferometer are given in appendix [ emvcz ] .
in the previous section we derived a vc - z relation , equation ( [ eq : vcz3d ] ) , in which the visibility matrix is determined from the brightness matrix .
it is well known that the original vc - z theorem ( far - zone form ) for narrow - fields states that there is a two - dimensional fourier transform relationship between visibility and brightness . in astronomical interferometry the transform aspect of this relationship is exploited to produce brightness images from measured visibility .
the wide - field vc - z , equation ( [ eq : vcz3d ] ) , is not a two - dimensional fourier transform , but , as we will now show , it is still possible invert it and thereby establish a sort of generalised transform .
first we should state that there are several ways of expressing @xmath93 in terms of @xmath53 , even though the wide - field vc - z relation , equation ( [ eq : vcz3d ] ) , is a one - to - one relationship in general .
this is because @xmath53 has redundancies , as one can expect considering the asymmetry in the respective matrix dimensions of @xmath93 and @xmath53 .
so although the rank of @xmath53 is three in general , it is overdetermined if @xmath93 is known .
we will first derive a solution that is valid for the entire celestial sphere based on the full @xmath53 .
the case when only a projection of @xmath53 is available will be discussed in section [ sec : dual - pol - int ] .
consider that we are given the full @xmath110 and that we would like to solve equation ( [ eq : vcz3d ] ) for @xmath93 . by extending the solution of the scalar problem described in @xcite to the three - dimensional matrix relationship in equation ( [ eq : vcz3d ] ) we find that one approximate solution is @xmath111 } \ , \mathrm{d}u\mathrm{d}v\mathrm{d}w \mathrm{d}n .
\label{eq : vcz3dinv}\end{aligned}\ ] ] where @xmath112 is the @xmath77 space spanned by @xmath113 for @xmath13 and @xmath114 . from @xmath42
, we can find a solution for the two - dimensional brightness matrix , @xmath115 an analogous expression applies for @xmath116 but with the integration over @xmath117 running from @xmath75 to @xmath118 rather than @xmath118 to @xmath119 . for @xmath104 ,
an expression can be obtained by using the @xmath105 parametrised vc - z mentioned in the previous section .
we can now state the generalised transform as @xmath120 where @xmath121 reads `` is wide - field , polarimetric vc - z related to '' .
the relation from brightness matrix to visibility matrix is given ( for @xmath95 ) by equation ( [ eq : vcz3d ] ) , and the relation from visibility matrix to brightness matrix is given by equations ( [ eq : bri2frombri3 ] ) and ( [ eq : vcz3dinv ] ) .
note that the @xmath121 relation is not simply a two - dimensional fourier transform as in the narrow - field case .
furthermore , the difference in the dimensionality of @xmath53 and @xmath93 make it clear that equation ( [ eq : vcz3drel ] ) can not simply be a matrix generalisation of the scalar vc - z relation , as is sometimes assumed .
in practice it is common to use stokes parameters to characterise the brightness and visibility matrices over narrow fields . let us see how stokes parameters can be applied to the wide - field vc - z relation , equation ( [ eq : vcz3d ] ) .
let us first consider brightnesses .
the difference between the two - dimensional brightness matrix @xmath93 in equation ( [ eq : vcz3d ] ) and the usual two - dimensional brightness matrix in the paraxial approximation is that the former is defined on a spherical domain while the latter is defined on a cartesian plane .
locally , for a source at some @xmath91 , the two brightness matrices can be made equal .
thus we can define the stokes parameters in terms of the components of @xmath93 with respect to a spherical basis @xmath59 in analogy with cartesian basis as @xmath122 when the spherical system is aligned as it is in section [ sec : wolfvcz ] such that the intersection of its central meridian and its equator is located at the centre of the fov and the pole @xmath62 is towards the earth s north pole , then in a sufficiently narrow field around the centre of the fov the stokes brightnesses @xmath123 are approximately equal to the stokes parameters of the iau @xcite . if one wishes to conform with iau stokes parameters over the entire celestial sphere , then one can use equation ( [ eq : stokesbright ] ) with @xmath86 replaced by @xmath124 , equation ( [ eq : sph2cart_gen ] ) , where @xmath125 is the declination of the centre of the fov .
the correspondence is then that @xmath126 is the iau s @xmath127 , and @xmath128 is the iau s @xmath129 .
note however that the iau basis set for polarimetry is not the same as the cartesian basis set widely used in radio interferometry for defining source directions and baselines , which in this paper is denoted @xmath69 . in adopting both these systems , therefore
, a sacrifice must be made , and we have chosen to slightly modify the usual pauli matrices that are used to relate the brightness matrix to the stokes parameters . explicitly , the brightness matrix in terms of the stokes parameters in equation ( [ eq : stokesbright ] ) is in this paper @xmath130 as can be seen , the expansion of this expression into unitary matrices , as in ( * ? ? ?
6 ) , leads to matrices equivalent to pauli matrices but with a change of overall sign for the matrices associated with @xmath131 and @xmath132 . in terms of equation ( [ eq : stokesbright ] ) , we can write the three - dimensional brightness matrix in equation ( [ eq : vcz3d ] ) as @xmath133 where @xmath134 it is easy to see that the @xmath135 for @xmath136 depend only on @xmath137 , and that they become the three - dimensional analogs of the equivalent pauli matrices in equation ( [ eq : britwoinstokes ] ) . equation ( [ eq : b3instokes ] ) shows that for every look - direction @xmath91 , the three - dimensional brightness matrix has four degrees of freedom , here expressed as the four stokes parameters . in other words ,
the four stokes parameters completely characterise the partially polarized brightness also for wide fields , under the assumptions made in the derivation of equation ( [ eq : vcz3d ] ) ) . now let consider if stokes visibilities
can be extended to wide fields .
a consequence of the three - dimensionality of @xmath53 in equation ( [ eq : vcz3d ] ) is , however , that the stokes parameters can not in general fully characterise the electric visibility .
this is because , in contrast to @xmath42 , which has a rank of at most two due to the transversality condition ) implies that @xmath42 has rank two comes from the rank - nullity theorem of linear algebra , since equation ( [ eq : b3constr ] ) implies that dimension of the null space is one and the matrix dimension of @xmath42 is three , so @xmath138 is the rank of @xmath42 . ] , equation ( [ eq : b3constr ] ) , @xmath53 has no similar constraint .
indeed one can convince oneself of the full rank of @xmath53 by considering two distinct point sources : the weighted sum of their @xmath42 matrices at some @xmath139 point according to equation ( [ eq : vcz3d ] ) , will in general be rank three .
as there are only four stokes parameters , albeit complex - valued in the case of visibilities , they can not fully parametrise a rank three matrix .
alternatively , rather than use the standard four stokes parameters , one could also use complexified versions of the nine , real , generalised stokes parameters @xcite .
these parameters are analogous to the standard stokes parameters but can completely describe the coherence of the full three - dimensional electric field . in light of the discussion above , these generalised stokes parameters are particularly suitable for parametrising @xmath53 , but we will not discuss them here any further .
the vc - z theorem is a basic , fundamental physical relationship is independent of technology .
the measurement equation ( m.e . ) of radio astronomy , on the other hand , includes practical aspects of telescope measurements , in particular the instrumental response of the telescope .
usually it is a relationship between a 2@xmath522 brightness matrix and a 2@xmath522 cross - correlation matrix of the output - voltage of a dual - polarized interferometer . since in the past such two - dimensional m.e .
have been tacitly based on the paraxial approximation valid only for narrow fields , it is important to verify that it can be recovered from the wide - field , polarimetric vc - z , equation ( [ eq : vcz3d ] ) , for which the paraxial approximation is not used . although it is possible to do this in a simple , straightforward way , we choose to do it in a more detailed way , introducing a formalism that extends the usual two - dimensional , electric field based model of radio astronomical antenna response . to obtain a 2-d m.e .
we must first introduce a formalism for converting the full electric field to a voltage in the interferometer antenna .
an electric field @xmath14 at an antenna excites an open circuit voltage @xmath132 .
assuming linearity , these two quantities are related as @xmath140 where @xmath141 is the antenna effective length vector . in general
it is a function of incidence direction , i.e. @xmath91 , but here we will only use ideal , hertzian dipole antennas ( short electric dipoles ) .
these have the important property that their effective length does not depend on incidence direction , @xmath141 is just a constant unit vector , and so it directly samples the component of the electric field along its length .
if we have @xmath117 co - located antennas that have no mutual coupling , their output voltages can be written in a matrix form @xmath142 where the @xmath143-th row in the @xmath144 matrix contains the components of antenna effective length vector @xmath145 .
the matrix of antenna effective lengths , denoted @xmath146 , has physical dimension length and is a @xmath117-dimensional extension of the @xmath147 matrix in ( * ? ? ? * eq .
( 2 ) ) , for which @xmath148 and thus models dual - polarized antennas .
when @xmath149 and the antennas are linearly independent , then they sample the full three - dimensional electric field .
such an antenna system is called _ tri - polarized _ antenna in general , and tripole antenna if the three dipoles are approximately mutually orthogonal .
one can also include antennas that sample the magnetic field , and an arrangement of electric and magnetic antennas can be constructed so as to sample the full electromagnetic field at a point .
such antennas are called electromagnetic vector - sensors @xcite . in what follows
, we will only be interested in dual- or tri - polarized antenna systems . in particular ,
let us consider a dual - polarized antenna that consists of two co - located , non - mutually coupled dipole antennas , one aligned along @xmath150 and the other along @xmath71 .
the response of such a dual - polarized antenna is @xmath151 where @xmath152 is the antenna effective length .
we have changed the subscripts on the voltages to reflect the right - hand side of the equation , in other words , for this dual - polarized antenna , each voltage component is directly proportional to a unique cartesian component of the electric field regardless of the radiations incidence angle .
because of this property , it can be regarded as an ideal dual - polarized polarimeter element . associated with each dual - polarized antenna element
is a two - dimensional plane in which the polarization is defined and measured .
if this plane is the same for all of the elements in a dual - polarized interferometer or if all the planes are mutually parallel ( common design goal for polarimetric interferometers ) we will say that such an interferometer is _ plane - polarized _ , in analogy with plane - polarized waves .
note that a plane - polarized interferometer is not necessarily a co - planar interferometer , and that a non - plane - polarized interferometer may be co - planar .
if the plane of a plane - polarized interferometer is to be specified explicitly , we will say , e.g. in the case of equation ( [ eq : ideal2dpolmeter ] ) , that it is @xmath153-polarized .
although most existing polarimetric interferometers in radio astronomy are intended to be plane - polarized , it is possible to have more general antenna elements such as electromagnetic vector - sensor arrays or electric tripole arrays .
an real - life example of the latter is the lois test station , see @xcite .
the prime motivation for such a tri - polarized system is that it samples the full electric field in a single telescope pointing rather than just a projection .
now that we have introduced a model formalism for antenna response we can derive the 2-d m.e .
for the special but important case of the @xmath153-polarized ( hertzian dipole ) interferometer , that is , the polarization plane is normal to the centre of the fov .
the output voltages from the @xmath153-polarized elements located at points @xmath154 and @xmath155 are cross - correlated and the result can be expressed as the correlation matrix @xmath156 for @xmath157 , where the arguments 1 and 2 refer to baseline points @xmath154 and @xmath155 .
@xmath146 is matrix multiplied from the left and its transpose from the right with @xmath42 , but since it does not depend on @xmath91 in this case , it can be pulled out of the integral in ( [ eq : vcz3d ] ) , and so the correlator output can be written @xmath158 where @xmath159 is the @xmath153-polarized antenna elements , effective length matrix .
as one can see , the effect of @xmath160 is equivalent to projecting the field vectors into the @xmath153-plane and multiplying by @xmath152 . in terms of the brightness matrix
the correlator output is @xmath161}}{\sqrt{1-l^{2}-m^{2 } } } \,\mathrm{d}l\mathrm{d}m \label{eq : meuv}\end{aligned}\ ] ] where we have introduced the @xmath153-projected transformation matrix @xmath162 to simplify the final result , which is now clearly two - dimensional .
equation ( [ eq : meuv ] ) is a wide - field m.e .
but with the novel jones matrix @xmath163 that physically represents a projection of the three - dimensional electric field vector onto the @xmath153-plane . in the narrow fov limit @xmath164 ,
so we can approximate @xmath163 in equation ( [ eq : meuv ] ) up to first order in @xmath99 and @xmath100 by the two - dimensional unity matrix , and so @xmath165 this is the basic m.e . of astronomical interferometry ,
see ( * ? ? ?
* eq . ( 14.7 ) ) .
thus we have shown that the wide - field , polarimetric vc - z , equation ( [ eq : vcz3d ] ) indeed reduces to the usual two - dimensional , jones vector based m.e . in the paraxial limit .
the result , equation ( [ eq : meparaxial ] ) , depended on the particular the spherical coordinate system used as default in this paper .
only this particular choice reduces directly to the standard 2-d m.e . in the paraxial limit .
to see what the dual - polarized m.e .
equation ( [ eq : meparaxial ] ) misses by not including the third dimension along @xmath166 , let us go back to equation ( [ eq : vcz3d ] ) and let assume the paraxial approximation , @xmath164 . in this case @xmath167 where we have kept only terms of first order in @xmath99 and @xmath100 .
@xmath168 in equation ( [ eq : vcz3d ] ) is identical to @xmath169 in equation ( [ eq : meparaxial ] ) for @xmath170 , but for the @xmath166 components , @xmath171 and @xmath172 so the @xmath173 component of the three - dimensional visibility matrix does not provide anything , but the @xmath174 ( and @xmath175 ) and the @xmath176 ( and @xmath177 ) components do .
thus even for a narrow fov , a dual - polarized interferometer does not measure the full set of generally non - zero electric visibilities .
note that if the antenna array is not exactly plane - polarized , these additional visibility components will contribute to the output - voltage of such an array .
this leads to the important question of how serious the loss of visibility information is in a dual - polarized interferometer .
more specifically , we ask whether a plane - polarized interferometer , given by some @xmath178 matrix @xmath146 , can in general perform full polarimetry of an arbitrary source distribution .
the answer is that the source brightness matrix in some direction can be determined fully only if @xmath179 since in this case @xmath180 is invertible .
for the special but important case @xmath181 this condition is equivalent to @xmath182 thus , an @xmath153-polarized interferometer can recover the full polarimetry except on the great circle orthogonal to @xmath7 , which we may call the imaging horizon .
so , in wide - field imaging with @xmath153-polarized interferometers the image horizon can not be measured with a single telescope pointing .
under noise - free conditions , this would pose little problem , but if we add in the effects of noise then the great circle of directions for which full polarimetry is not feasible broadens as a function of the signal - to - noise ratio .
although the discussion above was mainly focused on plane - polarized interferometers , the three - dimensional formalism developed here can also be applied to more general dual - polarized interferometers .
in particular it can model the situation when the polarization planes of dual - polarized antennas in an array are not all parallel .
since several large arrays of dual - polarized antenna based interferometers are currently being planned for , this would address the important question of whether such arrays should strive to be plane - polarized or whether they should purposely not be plane - polarized to minimize the inversion problems mentioned above .
we now show that the m.e . for the @xmath153-polarized , hertzian dipole interferometer , equation ( [ eq : meuv ] ) ,
exhibits distortions that depend on the look - direction , that is , the images based on these brightnesses contain polarization aberrations @xcite .
this agrees with the general understanding in observational radio astronomy that the polarimetry of a telescope is worse off - axis than on - axis .
say we have measured the @xmath37 correlation matrix @xmath183 .
we can not use the formal solution ( [ eq : vcz3dinv ] ) directly since it is for the 3@xmath523 visibility , and it is not clear how to obtain the remaining , unmeasured @xmath166 components from the @xmath184 and @xmath185 components of @xmath186 . on the other hand , for the scalar case , the formal solution for producing a synthesized image is the following scalar , wide - field imaging equation : @xmath187 } \mathrm{d}u\mathrm{d}v\mathrm{d}w\mathrm{d}n , \label{eq : sclwfimage}\ ] ] where @xmath188 is the scalar brightness and @xmath189 is the scalar visibility , see equation ( 13 ) in @xcite who call it the 3d method of synthesis imaging .
let us apply ( [ eq : sclwfimage ] ) to each component of @xmath186 as if it were a scalar , thus creating a matrix analogue of the scalar , wide - field imaging equation .
the resulting brightness matrix @xmath190 ( polarized image ) is @xmath191 } \mathrm{d}u\mathrm{d}v\mathrm{d}w\mathrm{d}n .
\label{eq : image2d}\ ] ] however , this is not the true brightness matrix because we see from equation ( [ eq : meuv ] ) that @xmath192 it follows that @xmath190 is actually the projection of the three - dimensional brightness matrix into the @xmath153-plane .
the stokes @xmath153-projected brightnesses @xmath193 are based directly on @xmath190 in analogy with equation ( [ eq : stokesbright ] ) , that is , @xmath194 the relationship between these stokes brightnesses and the true stokes brightnesses @xmath195 , which are based on @xmath93 can found by recasting equation ( [ eq : briprojrel ] ) as @xmath196 where @xmath197 @xmath198 is a mueller matrix that quantifies the distortion of true stokes brightnesses based on the imaging equation ( [ eq : image2d ] ) . for a very narrow fov , @xmath199 and
@xmath198 is approximately unity . in general , however , @xmath198 is not the unit matrix with the effect that the perceived stokes vector is a distortion of the true stokes vector .
thus , without further processing a plane - polarized interferometer of short dipoles will exhibit polarization aberrations over wide fields .
by contrast , a tripole array interferometer is , at least in theory , polarimetrically aberration - free over a wide field , since the scalar wide - field imaging equation ( [ eq : sclwfimage ] ) applied to the components of its visibility matrix as if they were scalars gives @xmath42 , which can be interpreted as the exact @xmath93 in cartesian ( rather than spherical ) components . examples of these wide - field distortions are displayed in fig .
[ fig : poldist4by4 ] .
it shows aberration effects for source distributions that are constant over the entire hemisphere , by which we mean that @xmath200 , so the brightnesses do not explicitly vary with direction . ] .
fortunately , these distortions can be compensated for in the image plane since @xmath198 is invertible and well - conditioned as long as @xmath201 is not close to one .
distortion of various polarized source distributions across the hemisphere for an @xmath153-polarized interferometer .
all distributions are such that the stokes brightnesses are constant , that is , they do not vary explicitly with direction although they may vary implicitly due to variation of the reference system for the stokes brightnesses , @xmath66 and @xmath68 , with direction .
the plots are of the polarization ellipses corresponding to the normalized stokes parameters @xmath202 , where @xmath203 are the @xmath153-projected stokes brightnesses as a function of the direction @xmath91 .
the plots clearly show polarization aberration , that is , a direction dependent distortion in the observed polarization .
the source distributions state of polarization ( shown in red ) can be seen at the centre of the fov , @xmath204 , where there is no distortion .
the upper - left panel shows @xmath205 , i.e. completely unpolarized radiation , the upper - right panel shows circularly polarized radiation , the lower - left shows radiation linearly polarized along @xmath66 , @xmath206 , and the lower - right panel is for radiation linearly polarized along @xmath68 , @xmath207 . ] another aspect of equation ( [ eq : stokespdist ] ) that is important , is that it shows , not surprisingly , that partially polarized , wide fields can not be treated as scalar , wide fields .
in fact , even for the supposedly ` scalar ' case , that is , when we only consider the scalar visibility @xmath208 of an unpolarized source distribution , so @xmath209 , equations ( [ eq : stokespdist ] ) and ( [ eq : image2d ] ) imply that @xmath210 where @xmath211 stands for two - dimensional fourier transform .
this should be compared with @xmath212 for the isotropic , scalar antenna case ( see , e.g. , equation ( 1 ) in @xcite ) .
this expression differs even for narrow fields since , to lowest , non - vanishing order in @xmath99 and @xmath100 , equation ( [ eq : myscalarvcz ] ) becomes approximately @xmath213 , \ ] ] while @xmath214 for the isotropic , scalar antenna case .
thus short dipoles are less aberrated than isotropic , scalar antennas .
the conclusion here , that scalar theory is not sufficient for the description of the general vc - z relations , agrees with those detailed in the field of optical coherence , see e.g. @xcite .
also , equation ( [ eq : myscalarvcz ] ) corresponds to an analogous equation in @xcite .
consider an unpolarized point source at @xmath215 , so @xmath216 , and express @xmath217 as @xmath218 ( where @xmath219 is the angle between the point source position and @xmath7 , and is equivalent to @xmath220 in @xcite ) , then the intensity of the unpolarized point source , as measured by single - pixel telescope , is @xmath221 .
this corresponds to equation ( 47 ) in @xcite .
although the results will be different for other types of antennas , the hertzian dipole is an important special case as it is the simplest polarimetric antenna and they are directly proportional to the cartesian coordinates of the electric visibility matrix .
the ultimate reason for the aberration is of course that the incident , transverse field is being projected onto the polarization plane of the hertzian dipoles thereby distorting the field . indeed
this projection is identical to the orthographic projection of a hemisphere in cartography . in comparison to other effects encountered in instrumental calibration , such as beam - shape
, these effects may seem small .
they are , however , important in that they determine the ultimate limits of polarimetry since these effects are of a intrinsic , geometric nature .
inspection of @xmath198 reveals that around @xmath204 the aberrations are @xmath222 . in the case of the aperture array of the ska
, the preliminary specification @xcite has a mean fov of 250 square degrees , so the aberration error is in the order of ~2% , or -16 db , at the edge of the fov .
this should be compared with the requirements from the key scientific projects that in some cases call for -30 db in polarization purity over wide fields .
thus , these polarization aberrations will need to be considered in wide - field imaging with ska .
it is not difficult to show that these polarization aberrations occurring at the edges of wide - field images also occur for off - axis imaging with fixed mount telescopes .
this occurs , for instance , in phased arrays of crossed dipoles such as lofar and the low frequency part of ska , where imaging at scan angles away from boresight ( zenith ) will be achieved by electronically steering the beam - form .
since the crossed dipoles are fixed to the ground and are not on a mechanically rotating mount , a situation geometrically analogous to the wide - field imaging consider above occurs .
this leads to similar aberrations in polarization for short electric dipole arrays .
scan angles of 45@xmath223 have been proposed for which the aberrations will of the order of -3 db .
we have derived the full set of electromagnetic vc - z relations , which are the basis of radio astronomical interferometry , without invoking the paraxial approximation .
these relations allow all - sky imaging in a single telescope pointing .
we have achieved these relations by generalising the usual 2-d cartesian jones vector based m.e . to a 3-d wolf coherence matrix formulation on the ( celestial ) sphere .
the derived wide - field vc - z relations are not simply trivial matrix ( or vector ) analogues of the wide - field , scalar vc - z , they exhibit direction dependent polarimetric effects . indeed even in the scalar limit ( that is , unpolarized radiation ) , our m.e . is not the same as the m.e . derived from scalar theory , in the case of hertzian dipoles .
furthermore , we found that , for an arbitrary wide - field of sources , the electric visibilities generally have nine complex components for an arbitrary baseline .
this implies that the standard stokes ( cartesian ) visibilities do not provide a full description of electric coherence for wide fields .
we have also shown that our vc - z relation ( for the electric field ) reduces to the standard 2-d m.e .
after a 2-d projection .
we showed that a consequence of this projection is that plane - polarized , hertzian dipole interferometers are aberrated polarimetrically .
fortunately , these aberrations can be corrected for in the image plane for sources sufficiently far away from the plane of the dual - polarized antenna elements . besides its use in the derivation of the wide - field vc - z relation
, we believe that the 3-d wolf formalism can be useful in constructing more general 3-d m.e . in cases requiring the full set of electric components . as examples ,
we mention the modelling of tri - polarized element arrays , the modelling of dual - polarized element arrays that are not strictly plane - polarized ( due to manufacturing errors or the earth s curvature ) , and the modelling of propagation that is not along the line - of - sight ( due to refraction or diffraction in the ionosphere , e.g. ) .
we thank the referee , j. p. hamaker , for valuable and constructive comments .
this work is supported by the european community framework programme 6 , square kilometre array design studies ( skads ) , contract no 011938 , and the science and technology facilities council ( stfc ) , uk .
bergman j. , carozzi t. d. , karlsson r. , 2003 , international patent publication , wo03/007422 bergman j. e. s. , hln l. , stl o. , thid b. , ananthakrishnan s. , wahlund j .- e .
, karlsson r. l. , puccio w. , carozzi t. d. , kale p. , 2005
, in dglr int .
symposium `` to moon and beyond '' , bremen , germany elvis - electromagnetic vector information sensor j. e. s. , carozzi t. d. , 2008 , preprint ( arxiv:0804.2092 ) bhatnagar s. , cornwell t. j. , golap k. , uson j. m. , 2008 , a&a , 487 , 419 brouw w. n. , 1971 , ph.d .
thesis , sterrenwacht leiden carozzi t. , karlsson r. , bergman j. , 2000 , phys .
e , 61 , 2024 carter w. h. , 1980 , j. opt .
, 70 , 1067 cornwell t. , golap k. , bhatnagar s. , 2005 , in ieee international conference on acoustics , speech , and signal processing , 5 , pp 861864 cornwell t. j. , perley r. a. , 1992 , a&a , 261 , 353 guthmann a. w. , thid b. , 2005 , aip conference proceedings , 745 , pp 770773 hamaker j. p. , bregman j. d. , 1996 , a&as , 117 , 161 hamaker j. p. , bregman j. d. , sault r. j. , 1996 , a&as , 117 , 137 hamaker j. p. , 2000 , a&as , 143 , 515 jouttenus t. , setl t. , kaivola m. , friberg a. t. , 2005 , phys .
e , 72 , 046611 the lofar team , july 2007 , astronnews , 12 ludwig a. c. , 1973 , ieee trans .
ant . & prop .
, 21 , 116 mcconnell d. , carretti e. , subrahmanyan r. , 2006 , apj , 131 , 648 mcguire j. p. j. , chipman r. a. , 1990 , j. opt .
soc . am . a , 7 , 1614 mandel l. , wolf e. , 1995 , optical coherence and quantum optics . cambridge university press morris d. , radhakrishnan v. , seielstad g. a. , 1964 , ap .
j. , 139 , 551 nehorai a. , paldi e. , 1991 , proc .
25th asilomar conf . on signals ,
syst . and comput . , pacific grove , ca , 566 piepmeier j. r. , simon h. k. , 2004 , ieee geoscience and remote sensing letters , 1 , 300 schilizzi r. t. , alexander p. , cordes j. m. , dewdney p. e. , ekers r. d. , faulkner a. j. , gaensler b. m. , hall p. j. , jonas j. l. , kellermann k. i. , 2007 , technical report , preliminary specifications for the square kilometre array .
ska program development office saastamoinen t. , tervo j. , turunen j. , 2003 , opt .
commun . , 221 , 257 sault r. j. , bock d. c .- j . , duncan a. r. , 1999 , a&as , 139 , 387 taylor r. , bredeson c. , dever j. , guram s. , deshpande a. , ghosh t. , momjian e. , salter c. , gibson s. , 2006 , national astronomy and ionosphere center newsletter , 39 , pp 13 thid b. , 2004 , in nilsson b. , fisherman l. , eds . , mathematical modelling of wave phenomena . vxj university press , 315 thompson a. r. , moran m. m. , swenson g. w. j. , 2001 , interferometry and synthesis in radio astronomy
. john wiley & sons , inc .
wolf e. , 1954 , nuovo cimento , 12 , 884
in the previous sections we considered only the electric field and its auto - correlation , but now we look at the full electromagnetic field . one motivation to do this is that , especially for low radio frequencies , it is possible to sample both the electric and the magnetic fields and thereby measure electromagnetic coherence fully , see @xcite , @xcite or @xcite .
such sensors have been deployed in some radio interferometers and will possibly provide unique and novel astronomical measurements . for completeness
then , we present the rest of the electromagnetic correlations analogous to the electric vc - z relation in equation ( [ eq : vcz3drel ] ) .
we found previously that the @xmath36 electric visibility matrix was related to the @xmath37 electric brightness matrix as @xmath224 where @xmath225 denotes the vc - z relationship given by equation ( [ eq : vcz3d ] ) .
note that we have now changed the notation of @xmath53 to @xmath226 and @xmath93 to @xmath227 to indicate that these quantities represent auto - correlations of visibility electric field and the brightness jones vectors , respectively .
a full electromagnetic vc - z relationship between brightnesses and visibilities requires also the auto - correlations of the magnetic field and the cross - correlation between the electric and magnetic fields , see @xcite .
we define the electromagnetic visibilities for baseline @xmath228 with respect to the phase reference direction @xmath50 as @xmath229 for @xmath41 , where @xmath230 is the magnetic field at @xmath231 .
the electromagnetic brightnesses in direction @xmath19 are defined as @xmath232 for @xmath233 , where @xmath234 is the magnetic field from the source distribution in direction @xmath19 .
the magnetic field associated with the visibilities can be derived by applying faraday s law , @xmath235 where @xmath236 is the impedance of free space , to the electric field used in derivation in section [ sec : wolfvcz ] .
the result is that the magnetic analogues of the electric vc - z expressions involve a magnetic jones vector @xmath237 that is related to the electric jones vector through @xmath238 where @xmath239 or in other words @xmath240 this says that the magnetic jones vector is directly determined from the electric jones vector , and so no other independent electromagnetic source coherence statistics exist ( in the far - field zone ) other than the @xmath37 electric brightness matrix . repeating the derivation in section [ sec : wolfvcz ] but for the magnetic field we find
we can write the rest of the vc - z relations as @xmath241 thus one can see @xmath242 as a jones - like matrix that switches between electric and magnetic coherencies .
these relations provide a complete description of the second - order coherence of the electromagnetic radiation .
so , for instance , one can easily compute the poynting visibility vector using the above relations , @xmath243\right\rangle = \nonumber \\ & \quad -\frac{1}{z_{0}}\iint i\mathbf{s } \mathrm{e}^{-\mathrm{i}2\pi\left[ul+vm+w\left(n -1\right ) \right ] } \ , \mathrm{d}\omega\end{aligned}\ ] ] where @xmath244 .
this says that the power flux visibility vector is vc - z related to the stokes brightness propagating from sources .
in this paper we have used the standard definitions of the @xmath77- and @xmath78-spaces , see @xcite , and we have used a spherical basis @xmath245 .
both these coordinate systems are relative to the centre of the fov .
often one wants to transform to some other system , and some details can be found in the standard texts @xcite .
however , what is not usually done explicitly is the transformation of the non - scalar brightnesses and visibilities .
we will now show how to rotate the matrix brightnesses and visibilities found in the polarimetric , wide - field vc - z , equation ( [ eq : vcz3d ] ) . equation ( [ eq : vcz3d ] ) uses the cartesian @xmath69 system as a basis for @xmath53 , @xmath19 , @xmath228 and the rows of @xmath86 .
it uses the spherical base vectors @xmath245 as a basis for @xmath93 and the columns of @xmath86 . under a rotation given by the 3@xmath523 orthogonal matrix @xmath246 ,
the vectors and matrices in the cartesian system transform in the usual manner : @xmath247 note that a transformation analogous to equation ( [ eq : vis3trans ] ) for the @xmath37 matrix @xmath248 , discussed in section [ sec : recover2dme ] , does not exist since @xmath248 lacks one of the dimensions necessary for a general coordinate transformation . only the correlator output from a tri - polarized antenna array
can be fully transformed in general .
actually , @xmath246 need not be the same in equations ( [ eq : vis3trans ] ) and ( [ eq : bri2trans ] ) , since the spherical and cartesian systems can be rotated separately
. this can be used to change the relative alignment ( rotation ) of the spherical system relative the cartesian system .
the net result is a change in matrix @xmath86 that relates the spherical components to the cartesian components .
consider the special case when the spherical system is rotated in the positive sense around the @xmath150-axis through angle @xmath85 relative the cartesian system .
the effect on the vc - z relations is that @xmath86 is replaced by @xmath87 , where @xmath251 obviously , for @xmath252 we obtain @xmath253 , which is the matrix used in most of this paper . at the field centre , @xmath254 , so @xmath255 for all @xmath256 .
however , the @xmath184 and @xmath185 components of its first derivatives are zero only for @xmath252 , that is @xmath257 only for @xmath252 .
thus , the special case we have used in this paper , @xmath253 , can be said to possess a projection ( @xmath153 ) that is locally flat at the field centre .
this is a reason for choosing the spherical system with @xmath252 as a default , since all other cases would lead to a 2-d m.e .
( [ eq : meparaxial ] ) with additional first - order terms that account for the coordinate system curvature within the fov .
the matrix @xmath87 can be used to adapt the vc - z relations given in this paper , such as equation ( [ eq : vcz3d ] ) , to standard celestial coordinate systems such as the equatorial system or the azimuth - elevation system .
in essence , assuming that the pole @xmath62 is towards the earth s north pole ( in the case of equatorial coordinates ) or zenith ( in the case of az - el coordinates ) , one simply performs @xmath258 and then interprets @xmath125 as either the declination ( equatorial case ) or the elevation ( az - el case ) of the centre of the fov .
note however that the cartesian coordinates , in which the @xmath77 and @xmath78 spaces are expressed , would still need to transformed for complete agreement with these standard celestial systems , but for this common task we refer to standard texts such as @xcite . | we derive a generalised van cittert - zernike ( vc - z ) theorem for radio astronomy that is valid for partially polarized sources over an arbitrarily wide field - of - view ( fov ) .
the classical vc - z theorem is the theoretical foundation of radio astronomical interferometry , and its application is the basis of interferometric imaging . existing generalised vc - z theorems in radio astronomy assume , however , either paraxiality ( narrow fov ) or scalar ( unpolarized ) sources .
our theorem uses neither of these assumptions , which are seldom fulfilled in practice in radio astronomy , and treats the full electromagnetic field . to handle
wide , partially polarized fields , we extend the two - dimensional electric field ( jones vector ) formalism of the standard `` measurement equation '' of radio astronomical interferometry to the full three - dimensional formalism developed in optical coherence theory .
the resulting vc - z theorem enables all - sky imaging in a single telescope pointing , and imaging using not only standard dual - polarized interferometers ( that measure 2-d electric fields ) , but also electric tripoles and electromagnetic vector - sensor interferometers .
we show that the standard 2-d measurement equation is easily obtained from our formalism in the case of dual - polarized antenna element interferometers .
we find , however , that such dual - polarized interferometers can have polarimetric aberrations at the edges of the fov that are often correctable .
our theorem is particularly relevant to proposed and recently developed wide fov interferometers such as lofar and ska , for which direction - dependent effects will be important .
[ firstpage ] telescopes ; techniques : interferometric ; techniques : polarimetric ; instrumentation : interferometers ; instrumentation : polarimeters . |
it is most essential and important issue that whether neutrino has nonzero mass and whether there is neutrino oscillation because which is related to the theoretical structure of standard model ( sm ) of particle physics and is also related to astrophysics and cosmology .
recent experiments@xcite-@xcite have shown that there are signals of neutrino oscillation .
if it is confirmed , neutrino system may not only be a window beyond sm but also be a fruitful system for testing quantum mechanics and probing discrete symmetries of nature as in @xmath4 system .
the suggestion that quantum coherence might be lost at the microscopic level was made by hawking@xcite , which suggested that asymptotic scattering should be described in term of a superscattering operator @xmath5 , relating initial and final density matrices , that does not factorize as a product of @xmath6 and @xmath7 matrix elements @xmath8 the loss of quantum coherence was thought to be a consequence of microscopic quantum- gravitational fluctuations in the space- time background .
ellis , hagelin , nanopoulos and srednicki ( ehns)@xcite then pointed out that if eq.(1 ) is correct for asymptotic scattering , there should be a corresponding effect in the quantum mechanics liouville equation that describes the time evolution of the density matrix @xmath9 @xmath10+i\delta \not{h}\rho\ ] ] the extra term may evolve a pure state into mixed state and result in quantum mechanics violation ( qmv ) .
the @xmath11 was parameterized for simple two level system by ehns@xcite and was used in neutron and @xmath4 systems .
then the qmv was through studied by the authors of references@xcite,@xcite,@xcite for @xmath4 system , and by the authors of ref.@xcite for @xmath12 system . some restrictions on qmv parameters
have been given .
a beam of two flavor neutrinos is a simple two - state system .
if mass eigenstate is not degeneracy with weak interaction eigenstate , then mixing between two flavor will exist and neutrino oscillation will occur . as in @xmath4 system ,
the quantum mechanics violation originate from quantum gravitational fluctuation may also exist in neutrino system .
we investigate qmv in neutrino system in term of the formalism proposed by ehns@xcite and find that neutrino oscillation probability can be modified by qmv effect .
thus the neutrino oscillation experiments may present a precise testing for quantum mechanics .
the main goal of this paper is that establish a framework of qmv in neutrino system , derive the detailed formulas of neutrino oscillation probability including qmv effect , and present some bounds on qmv parameters . for completeness
we will first derive the formulas of neutrino oscillation probabilities described by density matrix formalism within conventional quantum mechanics in section ii .
the qmv for neutrino oscillation in vacuum and in matter will be investigated in sections iii and iv respectively .
we will apply the formulas derived in previous section to current neutrino experiments and draw out certain restrictions on qmv parameters in section v. our conclusions are presented in section vi .
we now study a system consisted of two - flavor neutrinos in vacuum .
the hamiltonian of this system , in mass eignstate , is diagonal , @xmath13 , here @xmath14 and @xmath15 are energy eigenvalues of neutrinos . in some practical problems , for instance solar neutrino detection or atmospheric neutrino observation ,
@xmath16 are always valid , then @xmath17 , are good approximations , here @xmath18 is the magnitude of three momentum of neutrinos .
we describe the time evolution of the neutrino system by liouville equation in conventional quantum mechanics @xmath19,\ ] ] where @xmath20 is the density matrix of the system . following ehns[7 ] ,
we expand the hamiltonian and density matrix in pauli matrix basis @xmath21 where @xmath22 is unit matrix and @xmath23 are pauli matrices .
the evolution of the components of desity matrix obey @xmath24 we assume that there are only electron neutrinos in the system at the moment @xmath25 . considering this initial condition and solving the differential equation of eq.(5 )
, we get the density matrix @xmath26 where the @xmath27 is the vacuum mixing angle of neutrinos . the probability that electron neutrino oscillate to muon neutrino is given by computing the expection value of @xmath28 of an observable @xmath29 it is easy to check that the oscillation probability @xmath30\ ] ] is consistent with that obtained from solving propagation equation of neutrino@xcite .
if neutrino propagate in matter , the hamiltonian is corrected as@xcite @xmath31 where @xmath32 the @xmath33 and @xmath34 are the number densities of electrons and neutrons in matter .
the hamiltonian can be diagonalized by an unitary matrix @xmath35 , here the @xmath36 is effective mixing angle of neutrinos in matter , which is determined by @xmath37 the diagonalized hamiltonian can be expanded in pauli matrix basis as @xmath38 here @xmath39,\ ] ] @xmath40 in this situation , the @xmath41 in eq.(5 ) becomes @xmath42 in general , it is difficult to solve analytically the evolution equation of the components of density matrix , since the @xmath43 is a complicated function of time .
however , it is shown ( see appendex a ) that if adiabatic condition@xcite is valid , then the approximation @xmath44 is available , the term being proportional to @xmath45 can be neglected during solve the differential equation of density matrix
. the evolution equation of density matrix can be analytically solvable .
we still assume that there is only electron neutrino in system and the effective mixing angle is @xmath36 at the moment @xmath25 . under this initial condition
the desity matrix can be evaluated out @xmath46 if neutrinos are detected in vacuum , the oscillation probability is calculated as @xmath47 which is just the well known msw solution@xcite of solar neutrino problem . +
we now introduce qmv in neutrino oscillation .
following ehns@xcite the extra term in eq.(2 ) may be parameterized as a symmetrical @xmath48 matrix @xmath49 , then the evolution equation of the components of density matrix is rewritten as @xmath50 the matrix @xmath49 must obey some restrictions .
the probability is conserved , entropy must be real and never decrease , which imply that @xmath51 and @xmath49 is negative semidefinite . the energy conservation in neutrino oscillation demands that @xmath52 have to be imposed . the @xmath53 is thus parameterized as @xmath54 @xmath55 by solving the eq.(19 ) we get the density matrix @xmath56 where the @xmath57 and @xmath58 have different forms due to the magnitude relation between the difference of neutrino mass square and the qmv parameters , which will be discussed as follows . + (
i ) if the condition @xmath59 is satisfied , then @xmath20 has a oscillation - like solution @xmath60 @xmath61 where @xmath62 the oscillation probability is @xmath63 + ( ii ) .
if the relation @xmath64 is valid , the @xmath20 has an exponential - like solution @xmath65 e^{-\frac { \alpha+\gamma}{2 } t } , \ ] ] @xmath66 e^{-\frac{\alpha+\gamma}{2}t } , \ ] ] here @xmath67 the probability of neutrino oscillation becomes @xmath68 .\ ] ] it is easy to show that @xmath69 , the pure state may evolve into mixed state . with the @xmath70 , the @xmath71 return to eq.(6 ) in these two situations , quantum mechanics recovers .
+ in the situation of neutrino oscillation in matter , we add the parameterized @xmath72 term of eq.(20 ) into the hamiltonin of eq.(15 ) and get the evolution equation of the components of density matrix @xmath73 taking the adiabatic approximation and the initial condition that only electron neutrinos exist in the system into account , solving differential equations eq.(30 ) , we obtain the density matrix @xmath74 where @xmath36 is the effective mixing angle of neutrinos in matter at the initial moment .
the @xmath57 and @xmath58 are given as below .
+ ( i ) .
in the situation of @xmath75 then @xmath76 @xmath77 e^{-\frac{\alpha + \gamma}{2 } t } , \ ] ] where @xmath78 the neutrino oscillation probability is calculated as @xmath79 ( ii ) in another situation @xmath80 then @xmath81 @xmath82 e^{-\frac{\alpha + \gamma}{2 } t } , \ ] ] where @xmath83 the neutrino oscillation probability becomes @xmath84 it is not difficult to see that with the qvm parameter @xmath85 , the probability @xmath86 back to eq.(18 ) in the two situations , which is adiabatic msw solution @xcite of solar neutrino problem .
it is easy to calculate that @xmath87 . in this case
, pure state may evolve into mixed state , and thus quantum mechanics is violated .
we have derived the formulas of qmv effect in neutrino oscillation , which is to be tested in experiments .
a complete analysis requires a detailed understanding of all neutrino experiments which goes beyond the scope of this paper .
we present here only a rough estimate of maximum order of magnitude of the qmv parameters as an illuminating example of testing quantum mechanics in neutrino oscillation .
it is estimated theoretically that the maximum possible order of magnitude for qmv parameters @xmath88 , @xmath89 , or @xmath90 is @xmath91@xcite , which in @xmath92 system is @xmath93 , where @xmath94 is a typical energy scale in the system under discussion , and @xmath95 is the plank energy scale . the bounds on these parameters from @xmath92 experiments have been obtained in refs.@xcite,@xcite and @xcite , the last one gives @xmath96 similary , in neutrino system , it is expected theoretically that the order of magnitude of the maximum parameter is @xmath97 , where @xmath98 is a typical energy scale of neutrino system .
this is the order of @xmath99 for solar neutrinos or reactor neutrinos . in general
, we may reasonably assume that the order of the maximum parameters in neutrino system is @xmath100 , and that the magnitude relation among these parameters is retained in neutrino system , i.e. , @xmath101 . in the formulas of neutrino oscillation probability , the qmv parameters appear always in company with time @xmath102 , for example @xmath103 .
there is no enough large time to have the @xmath104 in current territorial experiments .
therefore , we discuss here only solar neutrino experiments because of large neutrino propagation time .
for the situation of neutrino oscillation in vacuum which is a possible solution of solar neutrino problem@xcite , from eq.(25 ) , we get @xmath105 where ( and hereafter ) the @xmath106 is simply written as @xmath107 .
if the condition eq.(26 ) is satisfied , we get @xmath108 > from above arguments of the order of magnitude of the maximum parameters and the order of maximum parameter of @xmath109 system in eq.(40 ) , we get an upper limit of another allowed region @xmath110 two allowed regions of the @xmath111 are decided by eq.(41)-(43 ) .
we studied the upper limit of @xmath112 given by eq.(41 ) vary with the @xmath113 , the numerical results is shown in fig.1 .
the lower and upper limits of the second allowed region are also shown in same figure .
the neutrino oscillation probability @xmath107 is taken from the kamiokande@xcite experimental data and the prediction of standard solar model ( ssm)@xcite . the region of @xmath114 is given by recent fit@xcite of world solar neutrino experiments .
the @xmath115 is taken as the center value of @xmath116 , and @xmath117 is used .
in other situation , we discuss neutrino oscillation in matter , which is another possible solution ( msw effect)@xcite of solar neutrino problem .
> from eq.(35 ) we get @xmath118 > from eq.(37 ) and eq.(21 ) , we get @xmath119 two allowed regions of @xmath111 are given by eqs.(43)-(45 ) for the case of including matter effect .
> from the kamiokande experimental data , ssm predictions , as well as the fit@xcite of experimental data to the adiabatic msw solution of solar neutrino problem , we obtain the numerical result of variation of the parameter @xmath111 with the @xmath120 , which is shown in fig.2 . in numerical calculation
, we used the center value of the @xmath121 given in ref.@xcite , @xmath122 , and @xmath117 .
we also apply the average value of the @xmath123 over the radius of the sun .
the electron distribution in the sun is taken from ref.@xcite .
it should be emphasized that our numerical result is only an example far from an exact analysis .
the aim is to show how the information of bounds on qmv parameters is extracted from neutrino oscillation experiments .
the value of the bounds are different if the different experimental data are used .
a complete numerical analysis based on our analytical formulas is necessary .
it is noted that the qmv for neutrino system in vacuum was discussed in ref.@xcite .
however , there is no analytical result and no any information about bounds on parameters in their disscussions .
moreover , what discussion in ref.@xcite is the case of neutrino oscillation in vacuum , but the neutrino mixing parameters used in their calculation are taken form the values of msw effect given in ref.@xcite , there is no any room of small mixing angle to be allowed by experiments for vacuum oscillation solution of solar neutrino problem .
neutrinos may be a useful system to precisely test quantum mechanics .
the liouville equation that describes the time evolution of density matrix of neutrino system can be modified by adding an extra term , which may be parameterized as a @xmath48 matrix .
the modified evolution equation of density matrix can be solved analytically if neutrinos propagate in vacuum or if neutrinos propagate in non - uniform matter but adiabatic condition is available .
the analytical expressions of neutrino oscillation probability with or without qmv effect have been derived .
based on theoretical analysis and neutrino experiment results we have extracted the bounds on the qmv parameters .
two allowed region of @xmath111 have been obtained .
it is expected that more precise restrictions on qmv parameters may be obtained from future neutrino experiments and complete numerical analysis . in this appendix
we will demonstrate that if the adiabatic condition is valid the term being proportional to @xmath124 can be neglected when we solve the evolution equation of the components of density matrix .
> from the eq(14 ) we get @xmath125 we define two dimensionless parameters @xmath126 and @xmath127 as @xmath128 where @xmath94 is a typical enegy scale in the system under discussion , which is taken as neutrino energy in the system considered here .
@xmath129 where @xmath130 are the eigenvalues of the matrix @xmath131 defined in eq.(10 ) .
the adiabatic condition is expressed @xcite as @xmath132 . from eqs.(11 ) and ( a1)-a3 ) , we get @xmath133 for the problems related to current neutrino experiments , we estimate the maximum value of the righthanded of eq.(a4 ) .
@xmath134 , @xmath135 , and @xmath136 , we thus get @xmath137 if adiabatic condition is valid , @xmath138 , we have @xmath139 .
therefore , we get the generalized adiabatic conditions in density matrix description of neutrino system
. we can safely neglected the term containing @xmath45 during we solve the differential equation of density matrix as long as the adiabatic condition is satisfied . under adiabatic approximation
the analytical solution of the evolution equation can be obtained .
j.n.bahcall , nucl . phys . * b*(proc . supll . )
* 48 * , 309 ( 1996 ) ; b.t.cleveland et al . ,
phys . * b*(proc . supll . )
* 38 * , 47 ( 1995 ) .
kamiokade ii collaboration , k.s.hirate et al .
* 65 * , 1297 ( 1990 ) ; * 65 * , 1301 ( 1990 ) ; * 66 * , 9 ( 1991 ) ; phys . rev . * d44 * , 2241 ( 1991 ) ; kamiokande iii colaboration , y. fukuda et al .
* 77 * , 1683 ( 1996 ) .
sage collaboration , a.i.abazov et al .
lett . * 77 * , 4708 ( 1996 ) ; phys . lett.*b328 * , 234 ( 1994 ) ; gallex collaboration , p.anselmann et al .
, phys . lett .
* b285 * , 376 ( 1992 ) ; * b314 * , 455 ( 1993 ) ; * b327 * , 337 ( 1994 ) ; w.hampel et al . , ibid * b388 * , 384 ( 1996 ) .
imb collaboration , r.becker szendy et al .
, phys . rev .
* d46 * , 372 ( 1992 ) ; kamiokande collaboration , k.s.hirata et al . , phys
b280 * , 146 ( 1992 ) ; t. stanev , nucl . phys . * b*(proc . supll . ) * 48 * , 165 ( 1996 ) .
lsnd collaboration , c. athanassopoulos et al , phys .
lett . * 75 * , 2650 ( 1995 ) ; * 77 * , 3087 ( 1996 ) . s.hawking , commun .
phys . * 87 * , 395 ( 1982 ) .
j.ellis , s.hagelin , d.v.nanopoulos , and m.srednicki , nucl
b241 * , 381 ( 1984 ) .
j.ellis , n.e.mavromatos and d.v.nanopoulos , phys . lett . *
b293 * , 142 ( 1992 ) .
p.huet and m.e.peskin , nucl
b434 * , 3 ( 1995 ) .
j.ellis , n.e.mavromatos and d.v.nanopoulos , phys . rev . * d53 * , 3846 ( 1996 ) .
d.s.du , x.q.li , y.liu and x.h.meng , phys . rev . * d53 * , 2645 ( 1996 ) .
s.m.bilenky and s.t.petcov , rev .
phys . * 59 * , 671 ( 1987 ) .
t.k.kuo and j.pantaleone , rev .
phys . * 61 * , 937 ( 1989 ) ; p.b.pal report oits-470 ( 1991 ) .
h.bethe , phys . rev
56 * , 1305 ( 1986 ) ; v.barger , r.j.n.phillips , and k.whianant , phys . rev . *
d34 * , 980 ( 1986 ) .
l.wolfenstein , phys . rev . *
d17 * , 2369 ( 1978 ) ; s.p.mikheyev and a.yu.smirnov , sov . j. nucl .
* 42 * , 913(1985 ) ; sov .
jetp * 64 * , 4 ( 1986 ) .
p.i.krastev and s.t.petcov , phys . rev . * d53 * , 1665 ( 1996 ) .
j.n.bahcall and m.h.pinsonneaule , rev .
* 64 * , 885 ( 1992 ) .
n.hata and p.langacker , phys . rev . * d56 * , 6107 ( 1997 ) .
y.liu , l.hu and m.l.ge , phys .
rev.*d56 * , 6648 ( 1997 ) .
v.barger , n.g.deshpande , r.j.n.phillips , p.b.pal , and k.whisnant , phys . rev . *
d43 * , 1759 ( 1991 ) .
n.hata and p.langacker , phys . rev .
* d50 * , 632 ( 1994 ) .
fig.1 the limits of the parameters @xmath140 vary with the @xmath114 for neutrino oscillation in vacuum .
the curve iii is an upper limit for the case of @xmath141 .
the line ii is a lower limit for the case that above condition is not valid .
the line i is an upper limit from the theoretical arguements and the experimental upper limit of the maximum parameter in @xmath142 system .
the arrows indicate the two allowed regions . | a scenario of testing quantum mechanics in neutrino oscillation is presented .
the quantum mechanics violation(qmv ) that is motivated by arguments based on quantum gravity is investigated in neutrino system .
it is found that the evolution equation of density matrix including qmv effect is analytically resolvable for neutrino propagate in vacuum or in matter under adiabatic approximation .
the analytical formulas have been derived . some bounds on the related parameters
have been obtained from neutrino experiments .
-0.4 in = 6.2 in = 9.2 in -0.0 cm * testing quantum mechanics in neutrino oscillation * fengcai ma@xmath0 and haiming hu@xmath1 _
@xmath0 department of physics , liaoning university , shenyang 110036 , p.r.china_ _
@xmath2 ccast ( world laboratory ) , beijing 100080 , p.r.china_ _ @xmath3 institute of high energy physics , academia sinica , _ _ p.o.box 918 , beijing 100039 , p.r.china_ pacs
number(s ) : 14.60.pq , 03.65.bz |
consider the following multi - dimensional fractal burgers equation in @xmath3 : @xmath4 where @xmath5 and @xmath6 is a viscosity constant , and @xmath1 with @xmath7 is the usual fractional laplacian defined by @xmath8 this is a typical nonlinear partial integro - differential equation and is regarded as a simplified model for the classical navier stokes equation when @xmath9 .
recently , there has been great interest in studying the multi - dimensional burgers turbulence ( cf .
@xcite ) , the fractal burgers equation ( cf . @xcite ) and the fractal conservation law equation ( cf .
@xcite ) , etc .
all these works are based on the analytic approaches , especially the energy method , duhamel s formulation and the maximum principle .
the purpose of the present paper is to give a probabilistic treatment for a large class of quasi - linear partial integro - differential equations .
let us first introduce the main idea . by reversing the time variable
, one can write burgers equation ( [ el5 ] ) as the following equivalent backward form : @xmath10 now , consider the case of @xmath9 , and for a given smooth solution @xmath11 to the above equation , let @xmath12 solve the following stochastic differential equation ( abbreviated as sde ) : @xmath13,\ x_{t , t}(x)=x,\ ] ] where @xmath14 is a @xmath15-dimensional standard brownian motion on @xmath16 $ ] . by it s formula and the markov property of the solution
, it is well known that @xmath17 conversely , assume that @xmath18 solves the implicit system ( [ es1 ] ) and ( [ es2 ] ) ; then @xmath19 also solves the backward burgers equation ( [ es3 ] ) .
this type of implicit stochastic differential equation has been systematically studied by freidlin @xcite , chapter 5 ; see also @xcite .
let us now substitute ( [ es2 ] ) for ( [ es1 ] ) ; then @xmath20_{y = x_{t , s}(x)}\,\dif s+\sqrt { 2\nu}\,\dif w_s , \nonumber \\[-8pt ] \\[-8pt ] \eqntext{s\in[t,0],\ x_{t , t}(x)=x.}\end{aligned}\ ] ] as the markov property holds , one can write the above equation as a closed form , @xmath21,\ x_{t , t}(x)=x,\hspace*{-35pt}\ ] ] where @xmath22\}$ ] , and @xmath23 denotes the conditional expectation with respect to @xmath24 .
the question is this : suppose that the stochastic equation ( [ es4 ] ) admits a unique solution family @xmath25 .
does @xmath26 defined by ( [ es2 ] ) solve burgers equation ( [ es3 ] ) ? to answer this question ,
the key point is to establish the following markov property : for all @xmath27 and @xmath28 , @xmath29 so that equation ( [ es4 ] ) can be written back to ( [ ee2 ] ) .
this is not obvious since sde ( [ es4 ] ) involves a conditional expectation operator . on the other hand
, one can replace the brownian motion in equation ( [ es4 ] ) by an @xmath0-stable process , as is done in @xcite , so that we can give a probabilistic explanation for the burgers equation ( [ es3 ] ) .
basing on this simple observation , in this paper we are mainly concerned about the following general stochastic functional differential equation ( abbreviated as sfde ) driven by a lvy process @xmath30 : @xmath31 \\[-8pt ] \eqntext { s\in[t,0],\ x_{t , t}(x)=x,}\end{aligned}\ ] ] where @xmath32 , @xmath33 and @xmath34 are some lipschitz functionals ( see below ) . in section [ sec2 ]
, we are devoted to proving the existence and uniqueness of a short time solution as well as the markov property ( [ mar ] ) for equation ( [ sde0 ] ) under lipschitz assumptions on @xmath33 and @xmath34 .
moreover , a locally maximal solution is also achieved .
since lvy processes usually have poor integrability , we have to carefully treat the big jump part of lvy processes .
compared with the classical argument in freidlin @xcite , it seems that sfde ( [ sde0 ] ) is easier to handle since it is a closed equation .
next , in section [ sec3 ] we apply our results to a class of quasi - linear partial integro - differential equations ( abbreviated as pide ) and obtain the existence of short time solutions . here , we discuss two cases : @xmath33 and @xmath34 satisfy linear growth conditions , but lvy processes have finite moments of arbitrary orders ; @xmath33 and @xmath34 are bounded , but equation ( [ sde0 ] ) has a constant coefficient in the big jump part .
this is natural since only the big jump is related to the moment of lvy processes . in section [ sec4 ]
, we turn to the investigation of the following system of semi - linear pides ( nonlinear transport equation ) : @xmath35 where @xmath36 is the generator of the lvy process given by ( [ ge ] ) below .
it is observed that the following scalar conservation law equation can be written as the above form : @xmath37 in particular , the one - dimensional fractal burgers equation ( [ es3 ] ) takes the above form . in equation ( [ tr ] ) , since there are not any analytic properties to be imposed on @xmath36 , one can not appeal to the duhamel formula or the energy method to give an analytic treatment .
in this situation , the probabilistic approach seems to be quite suitable .
in fact , by using purely probabilistic argument , we shall prove in theorem [ main ] below that pide ( [ tr ] ) admits a unique maximal weak solution in the class of bounded lipschitz functions . in the nondegenerate case
( corresponding to the subcritical case for @xmath38 with @xmath2 $ ] ) , the existence of global solutions is also obtained by applying some gradient estimates .
we mention that for the one - dimensional burgers equation ( [ el5 ] ) , it has been proved in @xcite that the global analytic solution does exist for @xmath39 $ ] , and the finite time blow up solution also exists for @xmath40 .
however , in the critical case of @xmath41 , the existence of global solutions for the general equation ( [ tr ] ) is left open .
we conclude this introduction by introducing the following conventions : the letter @xmath42 with or without subscripts will denote a positive constant , whose value may change in different places . if we write @xmath43 , this means that @xmath44 depends only on these indicated arguments .
let @xmath45 be an @xmath46-valued lvy process on the real line and defined on some complete probability space @xmath47 , which means that : * @xmath45 has independent and stationary increments , that is , for all @xmath48 , the random variables @xmath49 are independent , and the distribution of @xmath50 does not depend on @xmath51 .
* for @xmath52-almost all @xmath53 , the mapping @xmath54 is right - continuous and has left - limit ( also called cdlg in french ) .
let @xmath55 be the total of all @xmath52-null sets . for @xmath56
, define @xmath57\}\vee\sn.\ ] ] by the independence of increments of the lvy process , it is easy to see that for @xmath58 , @xmath59 and @xmath60 are independent . for simplicity of notation , we write @xmath61 it is clear that @xmath62 if @xmath63 , and @xmath64 is left - continuous . moreover , @xmath65 is @xmath66-measurable . throughout this paper , we shall work on the negative time axes @xmath67 $ ] .
[ re1 ] for any measurable process @xmath68 , @xmath69 , by the predictable projection theorem ( cf .
@xcite , page 173 , theorem 5.3 ) , there always exists a predictable version of @xmath70 , which will be denoted by @xmath71 .
moreover , for any @xmath72 , by the regularization theorem of martingales ( cf .
@xcite , page 64 , proposition 2.7 and page 65 , theorem 2.9 ) , we have @xmath73 where the second equality follows by @xmath74 and a monotone class argument . by the lvy khintchine formula ( cf .
@xcite , page 109 , corollary 2.4.20 ) , the characteristic function of @xmath75 is given by @xmath76\nu(\dif z)\biggr]\biggr\ } \nonumber \\[-8pt ] \\[-8pt ] \nonumber & = : & e^{t\psi(\xi)},\end{aligned}\ ] ] where @xmath77 is a complex - valued function called the symbol of @xmath30 , and @xmath78 , @xmath79 is a positive definite and symmetric matrix , @xmath80 is a lvy measure on @xmath46 , that is , @xmath81 and @xmath82 we call @xmath83 the characteristic triple of @xmath75 .
if @xmath84 and @xmath85 , where @xmath7 , then @xmath75 is the @xmath0-stable process with the lvy exponent @xmath86 , and its generator is the fractional laplacian @xmath1 by multiplying a constant @xmath87 . by the lvy it decomposition ( cf .
@xcite , page 108 , theorem 2.4.16 ) , @xmath75 can be written as @xmath88 where @xmath89 is a brownian motion with covariance matrix @xmath90 , @xmath91 is the poisson random point measure associated with @xmath30 given by @xmath92 and @xmath93 is the compensated random martingale measure . here ,
@xmath94 and @xmath95 are independent .
the generator of @xmath75 is given by @xmath96 \\[-8pt ] \nonumber & & { } + \int_{\mr ^m } \bigl[u(x+z)-u(x)-1_{|z|<1 } \partial _
i u(x)z_i\bigr]\nu(\dif z).\end{aligned}\ ] ] here and after , we use the usual convention for summation : the same index in a product will be summed automatically . in the following ,
we denote by @xmath97 the space of all cdlg functions from @xmath98 to @xmath3 , which is endowed with the locally uniform metric @xmath99 . notice that this metric is complete , but not separable . for given @xmath100 and cdlg function @xmath101\to\mr^d$ ] ,
we extend @xmath102 to @xmath98 in a natural manner by putting @xmath103 for @xmath104 so that @xmath105 . in this subsection , we consider the following general sfde in @xmath3 driven by the lvy process @xmath106 : @xmath107}g_r(x_{t , r-},\me^{\sf_{r- } } ( \phi _ r(x_{t,\cdot } ) ) ) \,\dif l_r,\qquad t\leq s\leq0,\ ] ] where @xmath108 , @xmath109 is a measurable function , and @xmath110 is a uniformly lipschitz continuous functional in the sense that @xmath111 where @xmath112}|\omega(s)-\omega'(s)|)$ ] is the locally uniform metric on @xmath97 . the definition about the solutions to equation ( [ sde ] ) is given as follows : [ def ] for fixed @xmath100 and @xmath108 , an @xmath113-adapted cdlg stochastic process @xmath114 is called a solution of equation ( [ sde ] ) if for all @xmath115 $ ] , @xmath116}g_r(x_{r-},\me^{\sf_{r-}}(\phi_r(x_\cdot ) ) ) \,\dif l_r\qquad \mbox{a.s.}\ ] ] for @xmath117 , we say that equation ( [ sde ] ) is ( uniquely ) solvable on @xmath118 $ ] ( or @xmath119 $ ] ) if for all @xmath120 $ ] ( or @xmath121 $ ] ) and @xmath108 , equation ( [ sde ] ) has a ( unique ) solution starting from @xmath122 at time @xmath123 . in this definition
, it has been assumed that @xmath124 so that @xmath125 makes sense by remark [ re1 ] , and further the stochastic integral with respect to the lvy process in the definition makes sense .
below , we make the following assumptions on the coefficients and the l ' evy measure @xmath80 : for some @xmath126 and all @xmath69 , @xmath127 , @xmath128 , @xmath129 for some @xmath130 , @xmath131 condition ( * * h**@xmath132 ) , which is a restriction on the big jump of the lvy process , is equivalent to saying that the @xmath133-order moment of the lvy process is finite ; cf .
@xcite , theorem 25.3 .
it should be noticed that for @xmath0-stable process , condition ( * * h**@xmath132 ) is satisfied only for any @xmath134 .
now we prove the following result about the existence and uniqueness of solutions for equation ( [ sde ] ) in a short time .
[ th1 ] assume that and hold for some @xmath135 , and @xmath34 is a lipschitz continuous functional on @xmath97 ; see ( [ el1 ] ) .
then there exists a time @xmath136 such that equation ( [ sde ] ) is uniquely solvable on @xmath119 $ ] for any @xmath137-integrable initial value @xmath108 in the sense of definition [ def ] , and for some @xmath138 and any @xmath121 $ ] , @xmath139}|x_{t , s}(\xi)|^\beta\bigr)\leq c\me .\ ] ] moreover , if @xmath140 is nonrandom , then for any @xmath141 , the unique solution @xmath142 is @xmath24-measurable for all @xmath115 $ ] .
we prove the theorem for @xmath143 . for @xmath144 ,
the proof is similar and simpler .
fix @xmath100 , which will be determined below . for @xmath145 ,
set @xmath146 , and let @xmath147 be the picard iteration sequence defined as follows : for @xmath148 , @xmath149}g_r\bigl(x^{(n-1)}_{t , r-},\me^{\sf_{r- } } \bigl(\phi_r\bigl(x^{(n-1)}_{t,\cdot}\bigr)\bigr)\bigr)\,\dif l_r.\ ] ] set @xmath150 using the lvy it decomposition ( [ le - ito ] ) , one can write @xmath151 } \int_{|z|<1}\cg^{(n)}_r\cdot z\tilde n(\dif r,\dif z)\\ & & { } + \int_{(t , s ] } \int_{|z|\geq1}\cg^{(n)}_r\cdot z n(\dif r,\dif z)\\ & & { } + \int_{(t , s]}\cg^{(n)}_r\cdot b\,\dif r+\int_{(t , s]}\cg ^{(n)}_r\,\dif w^a_r\\ & = : & i^{(n)}_1(s)+i^{(n)}_2(s)+i^{(n)}_3(s)+i^{(n)}_4(s),\end{aligned}\ ] ] where @xmath152 by burkholder s inequality ( cf .
@xcite , theorem 23.12 ) and young s inequality , thanks to @xmath143 , we have that for any @xmath153 , @xmath154}\bigl|i^{(n)}_1(r)\bigr|^\beta\bigr)\\ & & \qquad\leq c \me\biggl(\int_{(t,0 ] } \int_{|z|<1}\bigl|\cg^{(n)}_r\cdot z\bigr|^2n(\dif r,\dif z)\biggr)^{\beta/2 } \\ & & \qquad\leq c\me\biggl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^{2-\beta}\int _ { ( t,0 ] } \int_{|z|<1}\bigl|\cg^{(n)}_r\bigr|^\beta \cdot|z|^2n(\dif r,\dif z)\biggr)^{\beta/2 } \\ & & \qquad\leq\varepsilon\me\bigl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta\bigr ) + c_\varepsilon\me\biggl(\int_{(t,0 ] } \int_{|z|<1}\bigl|\cg^{(n)}_r\bigr|^\beta \cdot|z|^2n(\dif r,\dif z)\biggr ) \\ & & \qquad=\varepsilon\me\bigl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta\bigr ) + c_\varepsilon\me\biggl(\int_{[t,0 ] } \int_{|z|<1}\bigl|\cg^{(n)}_r\bigr|^\beta \cdot|z|^2\nu(\dif z)\,\dif r\biggr ) \\ & & \qquad\leq\biggl(\varepsilon+c_\varepsilon|t|\int_{|z|<1}|z|^2\nu(\dif z)\biggr ) \me\bigl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta\bigr).\end{aligned}\ ] ] here and below , the constant @xmath42 or @xmath155 is independent of @xmath123 and @xmath156 . for @xmath157 , by it s formula , we have @xmath158}\bigl|i^{(n)}_2(r)\bigr|^\beta\bigr)\\ & & \qquad\leq\me \biggl(\int _ { ( t , s ] } \int_{|z|\geq1}\bigl|\bigl|i^{(n)}_2(r- ) + \cg^{(n)}_r\cdot z\bigr|^\beta-\bigl|i^{(n)}_2(r-)\bigr|^\beta\bigr|n(\dif r,\dif z)\biggr)\\ & & \qquad=\me\biggl(\int_{(t , s ] } \int_{|z|\geq1}\bigl|\bigl|i^{(n)}_2(r- ) + \cg^{(n)}_r\cdot z\bigr|^\beta-\bigl|i^{(n)}_2(r-)\bigr|^\beta\bigr|\nu(\dif z)\,\dif r\biggr)\\ & & \qquad\leq c \me\biggl(\int_{(t , s]}\bigl|i^{(n)}_2(r)\bigr|^\beta\,\dif r \biggr)+c \biggl(\int_{|z|\geq1}|z|^\beta\nu(\dif z)\biggr ) \me\biggl(\int_{(t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta\,\dif r\biggr),\end{aligned}\ ] ] which then implies that by ( * * h**@xmath132 ) and gronwall s inequality , @xmath159}\bigl|i^{(n)}_2(r)\bigr|^\beta\bigr)\leq c|t|\me\bigl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta\bigr).\ ] ] similarly , we have @xmath159 } \bigl|i^{(n)}_3(r)\bigr|^\beta\bigr)\leq ( |t|\cdot|b|)^\beta\me\bigl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta \bigr),\ ] ] and for any @xmath153 , @xmath159}\bigl|i^{(n)}_4(r)\bigr|^\beta\bigr)\leq ( \varepsilon + c_\varepsilon|t| ) \me\bigl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta\bigr).\ ] ] combining the above calculations , we obtain that for any @xmath153 , @xmath160}\bigl|z^{(n)}_{t , r}\bigr|^\beta\bigr)\leq ( \varepsilon + c_\varepsilon|t|)\cdot \me\bigl(\sup_{r\in[t,0]}\bigl|\cg^{(n)}_r\bigr|^\beta\bigr).\ ] ] noticing that by ( * * h**@xmath161 ) , @xmath162}\bigl|z^{(n-1)}_{t , s}\bigr|\bigr)\bigr),\ ] ] and in view of @xmath163 , we further have by doob s maximal inequality , @xmath164}\bigl|z^{(n)}_{t , s}\bigr|^\beta\bigr ) \leq(\varepsilon+c_\varepsilon|t|)c_0\me\bigl(\sup_{s\in [ t,0]}\bigl|z^{(n-1)}_{t , s}\bigr|^\beta\bigr).\ ] ] now , let us choose @xmath165 and then for all @xmath121 $ ] , @xmath166}\bigl|z^{(n)}_{t , s}\bigr|^\beta\bigr)\leq \frac{1}{2}\me\bigl(\sup_{s\in[t,0]}\bigl|z^{(n-1)}_{t , s}\bigr|^\beta \bigr)\leq \cdots\leq\frac{1}{2^n } \me\bigl(\sup_{s\in[t,0]}\bigl|z^{(0)}_{t , s}\bigr|^\beta\bigr).\ ] ] on the other hand , notice that @xmath167}g_r\bigl(x^{(1)}_{t , r-},\me ^{\sf _ { r-}}(\phi_r(\xi))\bigr)\,\dif l_r.\ ] ] as above , and using gronwall s inequality , it is easy to derive that @xmath168}\bigl|z^{(0)}_{t , s}\bigr|^\beta\bigr)\leq c\me cdlg stochastic process @xmath142 such that @xmath169}\bigl|x^{(n)}_{t , s}-x_{t , s}\bigr|^\beta \bigr)=0.\ ] ] by taking limits for equation ( [ ep3 ] ) , it is easy to see that @xmath142 solves sfde ( [ sde ] )
. moreover , estimate ( [ el6 ] ) follows from ( [ eo1 ] ) , ( [ el7 ] ) and ( [ el8 ] ) .
the uniqueness is clear from the above proof .
suppose now that @xmath170 is nonrandom .
from the picard iteration ( [ ep3 ] ) , one sees that for each @xmath148 and @xmath171 $ ] , @xmath147 is @xmath24-measurable . indeed , suppose that @xmath172 is @xmath24-measurable for each @xmath171 $ ] , and then it is clear that @xmath173 is independent of @xmath174 .
noticing that for @xmath175 , @xmath176 and @xmath177 is independent of @xmath178 , we have @xmath179 by induction method , starting from equation ( [ ep3 ] ) with @xmath170 , one finds that @xmath147 is also @xmath24-measurable for each @xmath171 $ ] .
so , the limit @xmath142 is also @xmath24-measurable . in this theorem ,
if @xmath180 does not depend on @xmath19 , then the short time solution can be extended to any large time by the usual time shift technique . in theorem [ th1
] , since we require @xmath163 , the result rules out the @xmath0-stable process with @xmath181 $ ] . in this subsection , we drop assumption ( * * h**@xmath132 ) in theorem [ th1 ] , and consider the following special form : @xmath182 } \int_{|z|<1}g_r(x_{t , r-},\me^{\sf _
{ r-}}(\phi _ r(x_{t,\cdot})))\cdot z\tilde n(\dif r,\dif z)\nonumber\\[-2pt ] & & { } + \int_{(t , s ] } \int_{|z|\geq1}z n(\dif r,\dif z ) \nonumber \\[-9pt ] \\[-9pt ] \nonumber & & { } + \int_{(t , s]}g_r(x_{t , r-},\me^{\sf_{r-}}(\phi_r(x_{t,\cdot } ) ) ) \cdot b\,\dif r \\[-2pt ] & & { } + \int_{(t , s]}g_r(x_{t , r-},\me^{\sf_{r-}}(\phi_r(x_{t,\cdot})))\,\dif w^a_r,\nonumber\end{aligned}\ ] ] where @xmath108 . in this equation
, the big jump part has a constant coefficient . in order to make sense for the integrals , we need to assume that @xmath33 and @xmath34 are bounded .
we have : [ th3 ] in addition to , we assume that @xmath33 is bounded , and @xmath34 is a bounded lipschitz continuous functional on @xmath97
. then there exists a time @xmath183 such that sfde ( [ sde2 ] ) is uniquely solvable on @xmath119 $ ] . moreover , if @xmath140 is nonrandom , then for any @xmath141 , the unique solution @xmath142 is @xmath24-measurable for all @xmath115 $ ] . for @xmath100 and @xmath108 , set @xmath184 , and
let @xmath147 be the picard iteration sequence defined as follows : @xmath185 } \int_{|z|\geq1}z n(\dif r,\dif z ) \nonumber\\[-2pt ] & & { } + \int_{(t , s ] } \int_{|z|<1}g_r\bigl(x^{(n-1)}_{t , r-},\me^{\sf_{r-}}\bigl(\phi _
r\bigl(x^{(n-1)}_{t,\cdot}\bigr)\bigr)\bigr)\cdot z\tilde n(\dif r,\dif z ) \nonumber \\[-9pt ] \\[-9pt ] \nonumber & & { } + \int_{(t , s]}g_r\bigl(x^{(n-1)}_{t , r-},\me^{\sf_{r-}}\bigl(\phi _
r\bigl(x^{(n-1)}_{t,\cdot}\bigr)\bigr)\bigr)\cdot b\,\dif r \\[-2pt ] & & { } + \int_{(t , s]}g_r\bigl(x^{(n-1)}_{t , r-},\me^{\sf_{r-}}\bigl(\phi _ r\bigl(x^{(n-1)}_{t,\cdot}\bigr)\bigr)\bigr)\,\dif w^a_r.\nonumber\end{aligned}\ ] ] set @xmath150 then @xmath186 } \int_{|z|<1}\cg^{(n)}_r\cdot z\tilde n(\dif r,\dif z ) + \int_{(t , s]}\cg^{(n)}_r\cdot b\,\dif r+\int_{(t , s]}\cg^{(n)}_r\,\dif w^a_r,\ ] ] where @xmath187 notice that @xmath188 , and by ( * * h**@xmath161 ) , @xmath189}\bigl|z^{(n-1)}_{t , s}\bigr|\bigr)\bigr)^2 \bigr)=:\phi^{(n)}_r.\ ] ] by burkholder s inequality and ( [ levy ] ) , we have @xmath190}\biggl|\int_{(t , s ] } \int_{|z|<1}\cg ^{(n)}_r\cdot z\tilde n(\dif r,\dif z)\biggr|^2\biggr ) & \leq & c\me\biggl(\int_{(t,0 ] } \int_{|z|<1}\bigl|\cg^{(n)}_r\cdot z\bigr|^2n(\dif r,\dif z)\biggr ) \\ & \leq & c|t|\me\bigl(\sup_{r\in[t,0]}\phi^{(n)}_r\bigr).\end{aligned}\ ] ] here and below , the constant @xmath42 is independent of @xmath123 and @xmath156 .
similarly , we have @xmath191}\biggl|\int_{(t , s]}\cg^{(n)}_r\cdot b\,\dif r\biggr|^2\biggr)\leq c|t|^2\me\bigl(\sup_{r\in[t,0]}\phi^{(n)}_r\bigr)\ ] ] and @xmath191}\biggl|\int_{(t , s]}\cg^{(n)}_r\,\dif w^a_r \biggr|^2\biggr)\leq c|t|\me\bigl(\sup_{r\in[t,0]}\phi^{(n)}_r\bigr).\ ] ] combining the above calculations and by doob s maximal inequality , we obtain @xmath164}\bigl|z^{(n)}_{t , s}\bigr|^2\bigr)\leq c_0|t|\me \biggl(\sup_{r\in[t,0]}\phi^{(n)}_r\biggr ) \leq c_1|t|\me\bigl(\sup_{s\in[t,0]}\bigl|z^{(n-1)}_{t , s}\bigr|^2\bigr).\ ] ] now , let us choose @xmath192 and then for all @xmath121 $ ] , @xmath159}\bigl|z^{(n)}_{t , r}\bigr|^2\bigr)\leq \frac{1}{2}\me\bigl(\sup_{r\in[t,0]}\bigl|z^{(n-1)}_{t , r}\bigr|^2 \bigr)\leq\cdots
\leq \frac{1}{2^n}\me\bigl(\sup_{r\in[t,0]}\bigl|z^{(1)}_{t , r}\bigr|^2 \bigr)\leq\frac{c}{2^n}.\ ] ] hence , there exists an ( @xmath193)-adapted and cdlg stochastic process @xmath142 such that @xmath194}\bigl|x^{(n)}_{t , s}-x_{t , s}\bigr|^2\bigr)=0.\ ] ] by taking limits for equation ( [ ep33 ] ) , it is easy to see that @xmath142 solves sfde ( [ sde2 ] ) .
the remaining proof is the same as in theorem [ th1 ] . in this subsection
, we prove the markov property for the solutions of equations ( [ sde ] ) and ( [ sde2 ] ) , which is crucial for the development of the next section .
we first show the continuous dependence of the solutions with respect to the initial values .
[ pro1 ] in the situation of theorem [ th1 ] , for @xmath121 $ ] , let @xmath195 .
if @xmath196 converges to @xmath122 in probability as @xmath197 , then @xmath147 converges to @xmath142 uniformly with respect to @xmath198 $ ] in probability as @xmath197 , where @xmath199 and @xmath200 are the solutions of sfde ( [ sde ] ) corresponding to the initial values @xmath196 and @xmath122 .
define @xmath201 then we can write @xmath202 } \int_{|z|<1}1_{a_n}\cg^{(n)}_r\cdot
z\tilde n(\dif r,\dif z)\\ & & { } + \int_{(t , s ] } \int_{|z|\geq1}1_{a_n}\cg^{(n)}_r\cdot z n(\dif r,\dif z)\\ & & { } + \int_{(t , s]}1_{a_n}\cg^{(n)}_r\cdot b\,\dif r+\int _ { ( t , s]}1_{a_n}\cg^{(n)}_r\,\dif w^a_r,\end{aligned}\ ] ] where @xmath203 as in ( [ eo1 ] ) , we can prove that for all @xmath121 $ ] , @xmath204}\bigl|x^{(n)}_{t , s}-x_{t , s}\bigr|^\beta \bigr)\leq c\me\bigl(1_{a_n}\bigl|\xi^{(n)}-\xi\bigr|^\beta\bigr),\ ] ] where @xmath42 is independent of @xmath156 .
now , for any @xmath205 , we have @xmath206}\bigl|x^{(n)}_{t , s}-x_{t , s}\bigr|\geq\eps\bigr\ } & \leq & p\bigl\{1_{a_n}\sup_{s\in[t,0]}\bigl|x^{(n)}_{t , s}-x_{t , s}\bigr|\geq\eps \bigr\ } + p(a^c_n)\\[-1pt ] & \leq&\frac{1}{\eps^\beta}\me\bigl(1_{a_n } \sup_{s\in [ t,0]}\bigl|x^{(n)}_{t , s}-x_{t , s}\bigr|^\beta\bigr)+p(a^c_n)\\ & \leq&\frac{c}{\eps^\beta}\me\bigl(1_{a_n}\bigl|\xi^{(n)}-\xi\bigr|^\beta\bigr)+p(a^c_n).\end{aligned}\ ] ] the proof is then complete by letting @xmath197 .
in the situation of theorem [ th3 ] , the conclusion of this proposition still holds , which can be proven by the same procedure .
the following lemma is a direct consequence of the uniqueness of solutions .
suppose that sfde ( [ sde ] ) is uniquely solvable on the time interval @xmath118 $ ] .
then for all @xmath207 and @xmath208 , we have @xmath209 moreover , for any @xmath210 , @xmath211 and disjoint @xmath212 with @xmath213 , @xmath214 for @xmath215 , we can write @xmath216}g_r(x_{t_1,r-}(\xi ) , \me ^{\sf_{r-}}(\phi(x_{t_1,\cdot}(\xi))))\,\dif l_r \qquad\mbox{a.s.}\ ] ] on the other hand , if we set @xmath217,\ ] ] then @xmath218 satisfies @xmath219}g_r(y_{r-},\me^{\sf_{r-}}(\phi ( y_\cdot)))\,\dif l_r\qquad \mbox{a.s.}\ ] ] equality ( [ ep1 ] ) follows by the uniqueness . as for ( [ ep2 ] ) , noticing that for all @xmath220 $ ] , @xmath221 & & \qquad = \sum_ig_r(1_{\lambda_i}x_{t , r-}(x_i),1_{\lambda_i}\me ^{\sf_{r- } } ( \phi_r(x_{t,\cdot}(x_i))))\vadjust{\goodbreak}\\ & & \qquad = g_r\biggl(\sum_i 1_{\lambda_i}x_{t , r-}(x_i),\me^{\sf_{r- } } \biggl(\phi_r\biggl(\sum_i 1_{\lambda_i}x_{t,\cdot}(x_i)\biggr)\biggr)\biggr),\end{aligned}\ ] ] it follows by the uniqueness as above .
now we can prove the following markov property .
[ pro2 ] in the situation of theorem [ th1 ] or theorem [ th3 ] , let @xmath222 be the solution family of sfde ( [ sde ] ) or ( [ sde2 ] ) . then for any @xmath223 , @xmath28 and bounded continuous function @xmath224 , we have @xmath225 we only prove ( [ es5 ] ) in the case of theorem [ th1 ] . by proposition [ pro1 ] ,
the mapping @xmath226 is continuous .
so,@xmath227 is @xmath228-measurable .
thus , for proving ( [ es5 ] ) , it suffices to prove that for any @xmath229 , @xmath230 let @xmath231 be a sequence of simple functions , where @xmath232 , @xmath233 disjoint and @xmath234 , and such that @xmath235 by proposition [ pro1 ] again , we have @xmath236 since @xmath237 is @xmath238-measurable and independent of @xmath228 , we further have @xmath239 the proof is complete . now , suppose that @xmath34 takes the following form : @xmath240 where @xmath241 and @xmath242 satisfy that for some @xmath243 and all @xmath244 and @xmath127 , @xmath245 assume that ( [ ep2 ] ) , and hold for some @xmath163 .
then there exists a time @xmath246 such that sfde ( [ sde ] ) is solvable on @xmath118 $ ] for any initial value @xmath28 , and if @xmath44 is finite , then @xmath247 where @xmath248 moreover , the family of solutions @xmath249 is unique in the class that for all @xmath207 and @xmath28 , @xmath250 we also have the following uniform estimate : for any @xmath251 and @xmath28 , @xmath252}\me\bigl(\sup_{s\in[t,0]}|x_{t , s}(x)|^\beta \bigr)\leq c_{t',x}.\ ] ] first of all , let @xmath253 be the existence time in theorem [ th1 ] .
by ( [ ep4 ] ) , there exists a constant @xmath254 such that for all @xmath127 and @xmath255 $ ] , @xmath164}|x_{t , s}(x)-x_{t , s}(x')|^\beta \bigr)\leq c|x - x'|^\beta.\ ] ] using this estimate and ( [ ep2 ] ) , it is easy to check that @xmath256 next , we consider the following sfde on @xmath257 $ ] : @xmath258}g_r\biggl(x_{t , r-}(x),\\ & & \hspace*{42pt}\qquad\me^{\sf _ { r-}}\biggl(u_{t_1}(x_{t , t_1}(x ) ) + \int^{t_1}_r f_{r'}(x_{t , r'}(x))\,\dif r'\biggr)\biggr)\,\dif l_r.\end{aligned}\ ] ] repeating the proof of theorem [ th1 ] , one can find another @xmath259 so that this sfde is uniquely solvable on @xmath260 $ ] .
meanwhile , one can patch up the solution by setting @xmath261,\ t\in[t_2,t_1].\ ] ] it is easy to verify that @xmath262 solves sfde ( [ sde ] ) on @xmath263 $ ] . proceeding this construction , we obtain a sequence of times @xmath264 and a family of solutions @xmath265 from the construction of @xmath44 , one knows that ( [ ep55 ] ) holds .
as for the uniqueness , it can be proved piecewisely on each @xmath266 $ ] .
estimate ( [ el66 ] ) follows from ( [ el6 ] ) and induction . by this theorem , for obtaining the global solution , it suffices to give an a priori estimate for @xmath267 .
the following result can be proved similarly .
we omit the details .
[ th22 ] in addition to ( [ ep2 ] ) and , we assume that @xmath268 and @xmath102 are uniformly bounded
. then there exists a time @xmath269 such that sfde ( [ sde2 ] ) is solvable on @xmath118 $ ] and estimate ( [ ep55 ] ) holds provided @xmath270 .
moreover , the family of solutions @xmath249 is unique in the class that for all @xmath207 and @xmath28 , @xmath271
in this section , we establish the connection between stochastic functional differential equations and a class of quasi - linear partial integro - differential equations . for this aim ,
we consider @xmath34 taking the form of ( [ es6 ] ) and assume that for some @xmath272 , ( * * h**@xmath273 ) , @xmath274 and @xmath224 are continuous functions in @xmath275 , and for any @xmath276 , @xmath277 , @xmath278 , @xmath279 are uniformly bounded continuous functions with respect to @xmath244 , where @xmath280 denotes the @xmath281th order gradient with respect to @xmath282 .
we also denote @xmath283 under this assumption , it is clear that ( [ ep2 ] ) and ( * * h**@xmath161 ) hold .
let @xmath26 be defined by ( [ ep3 ] ) . by theorem [ th2 ] ,
the mapping @xmath284 is lipschitz continuous .
however , it is in general not @xmath285-differentiable since we have poor integrabilities for @xmath286 .
we shall divide two cases to discuss this problem . in this subsection , we consider equation ( [ sde ] ) , and assume that ( * * h**@xmath273 ) holds for some @xmath287 , and ( * * h**@xmath132 ) holds for all @xmath144 . in this case
, we can write @xmath288 where @xmath289 . let @xmath117 be the maximal time given in theorem [ th2 ] and @xmath290 the solution family of equation ( [ sde ] ) .
for simplicity of notation , below we shall write @xmath291 \\[-9pt ] \nonumber & : = & g_r\biggl(x_{t , r-}(x),\me^{\sf _ { r-}}\biggl(\varphi(x_{t,0}(x ) ) + \int^0_r f_{r'}(x_{t , r'}(x))\,\dif r'\biggr)\biggr).\end{aligned}\ ] ] let @xmath292 be a @xmath285-function with bounded first and second order partial derivatives . by it s formula ( cf .
@xcite , page 226 , theorem 4.4.7 ) , we have @xmath293 } \int_{\mr^m}[g(x_{t , r-}+\cg _ { t , r}\cdot z)-g(x_{t , r- } ) \nonumber\\[-2pt ] & & \hspace*{100pt}\qquad{}-\partial _
i g(x_{t , r-})\cg^{ij}_{t , r } z_j]\nu(\dif z)\,\dif r \nonumber \\[-9pt ] \\[-9pt ] \nonumber & & { } + \int_{(t , s]}\partial _
i g(x_{t , r-})\cg^{ij}_{t , r } \hat b_j\,\dif r \\[-2pt ] & & { } + \frac{1}{2}\int_{(t , s]}\partial _
jg(x_{t , r-})(\cg ^{\mathrm { t}}_{t , r } a\cg _ { t , r})^{ij}\,\dif r+m^g_{t , s},\nonumber\end{aligned}\ ] ] where @xmath294 } \int_{\mr^m}[g(x_{t , r-}+\cg _ { t , r}\cdot z)-g(x_{t , r-})]\tilde n(\dif r,\dif z ) \\[-2pt ] & & { } + \int_{(t , s]}\partial _ ig(x_{t , r-})\cg^{ij}_{t , r } \,\dif(w^a_r)^j\end{aligned}\ ] ] is a square integrable ( @xmath24)-martingale by ( [ el66 ] ) . here and below , the superscript `` t '' denotes the transpose of a matrix .
fix @xmath120 $ ] and @xmath295 so that @xmath296 $ ] . by taking expectations for both sides of ( [ el3 ] ) , we have @xmath297=i^g_1(h)+i^g_2(h)+i^g_3(h),\ ] ] where @xmath298 & & \hspace*{92pt}\qquad{}-\partial _ i g(x_{t - h , r})\cg^{ij}_{t - h , r } z_j]\nu(\dif z)\,\dif r \biggr),\\[-2pt ] i^g_2(h)&:=&\frac{1}{h}\me\biggl(\int^t_{t - h}\partial _ ig(x_{t , r})\cg ^{ij}_{t - h , r } \hat b_j\,\dif r\biggr),\\[-2pt ] i^g_3(h)&:=&\frac{1}{2h}\me\biggl(\int^t_{t - h}\partial _ i\partial _ jg(x_{t , r})(\cg ^{\mathrm{t}}_{t - h , r } a\cg_{t - h , r})^{ij}\,\dif r\biggr).\end{aligned}\ ] ] we have : [ le2 ] as @xmath299 , it holds that @xmath300\nu ( \dif z),\\[-2pt ] i^g_2(h)&\to&\partial _
ig(x)\sg^{ij}_t(x)\hat b_j,\qquad i^g_3(h)\to \tfrac { 1}{2}\partial _ i\partial _ jg(x)(\sg^{\mathrm{t}}_t(x)a\sg_t(x))^{ij},\end{aligned}\ ] ] where @xmath301 we only prove the first limit , the others are analogous . by the change of variables ,
we can write @xmath302 & & \hspace*{48pt}{}-g(x_{t - h , t - hs } ) -\partial _ i g(x_{t - h , t - hs})\cg^{ij}_{t - h , t - hs } z_j ] \nu(\dif z)\,\dif s\biggr).\end{aligned}\ ] ] notice that @xmath303 & & \qquad=\int_{(t - h , t - hs ] } \int_{\mr^m}\cg _ { t - h , r}(x)\cdot z\tilde n(\dif r,\dif z ) + \int_{(t - h , t - hs]}\cg_{t - h , r}(x)\cdot\hat b\,\dif r\\[-2pt ] & & \qquad\quad { } + \int_{(t - h , t - hs]}\cg_{t - h , r}(x)\,\dif w^a_r\\[-2pt ] & & \qquad=:j_1(h)+j_2(h)+j_3(h).\end{aligned}\ ] ] by the isometric property of stochastic integrals , we have @xmath304 & \leq&|h|\me\biggl(\sup_{r\in[t - h , t - hs]}|\cg_{t - h , r}(x)|^2\int _ { \mr ^m}|z|^2\nu(\dif z)\biggr)\\[-2pt ] & \leq & c|h|\me\bigl(1+\sup_{r\in[t - h,0]}|x_{t - h , r}(x)|^2 \bigr)\stackrel { \scriptsize(\ref{el66})}{\to}0\qquad\mbox{as $ h\downarrow0$,}\end{aligned}\ ] ] where in the last inequality , we used doob s maximal inequality and that @xmath305 and @xmath102 in definition ( [ ep0 ] ) of @xmath306 are linear growth in @xmath307 and @xmath19 , respectively .
similarly , @xmath308 hence , for fixed @xmath309 , @xmath310 noticing that @xmath311 we have @xmath312\,\dif\theta\biggr)\\ & & \hspace*{245pt}{}\times \cg^{ij}_{t - h , t - hs } z_j\biggr|\\ & & \qquad\leq c\me|\cg_{t - h , t - hs}|^2 |z|^2\leq c\me\bigl(1+\sup_{r\in[t - h,0]}|x_{t - h , r}(x)|^2 \bigr)|z|^2\stackrel { \scriptsize(\ref{el66})}{\leq } c|z|^2,\end{aligned}\ ] ] where the second - to - last inequality is the same as above , and the constant @xmath42 is independent of @xmath313 .
thus , for proving the first limit , by the dominated convergence theorem , it suffices to prove that for fixed @xmath314 $ ] and @xmath315 , @xmath316\,\dif\theta\biggr)\cg^{ij}_{t - h , t - hs } z_j\biggr)\\ & & \qquad \to\biggl(\int^1_0\bigl[\partial _ i g\bigl(x+\theta\sg_t(x)\cdot z\bigr ) -\partial _ i g(x)\bigr]\,\dif\theta\biggr)\sg^{ij}_t(x ) z_j\qquad \mbox{as $ h\downarrow0$.}\end{aligned}\ ] ] by ( [ el4 ] ) and remark [ re1 ] , this limit is easily obtained .
we also need the following differentiability of the solution @xmath12 with respect to @xmath307 in the @xmath317-sense .
[ th4 ] for any @xmath318 , there exists a time @xmath319 , where @xmath320 is defined by ( [ ch ] ) , and @xmath321 is defined by ( [ kk ] ) , such that for any @xmath322 , the mapping @xmath323 is @xmath324-differentiable in the @xmath317-sense and for any @xmath325 , @xmath326}\me|\nabla^j x_{t , s}(x)|^p<+\infty.\ ] ] since the proof is standard ( cf .
@xcite , theorem 39 or @xcite , section 4.6 ) , we sketch it .
let @xmath327 be the canonical basis of @xmath3 .
for @xmath328 and @xmath329 , define @xmath330 and @xmath331 where @xmath332 is defined by ( [ ep0 ] ) .
then , @xmath333 } \int_{\mr^m}\cg^{\delta , i}_{t , r}\cdot z\tilde n(\dif r,\dif z ) + \int_{(t , s]}\cg^{\delta , i}_{t , r}\cdot\hat b\,\dif r+\int_{(t , s]}\cg ^{\delta , i}_{t , r}\,\dif w^a_r.\ ] ] as in ( [ ep5 ] ) , by burkholder s inequality , we have that for any @xmath318 , @xmath334}|x^{\delta , i}_{t , r}|^p\bigr)\leq c_{p,\sa}|t| \me\bigl(\sup_{r\in[t,0]}|\cg^{\delta , i}_{t , r}|^p\bigr).\ ] ] moreover , by ( * * h**@xmath273 ) and doob s maximal inequality , we easily derive that @xmath159}|\cg^{\delta , i}_{t , r}|^p\bigr ) \leq c_{p,\sk}\me\bigl(\sup_{r\in[t,0]}|x^{\delta , i}_{t , r}|^p\bigr).\ ] ] substituting this into ( [ ep1 ] ) , we find that for some @xmath335 independent of @xmath336 and @xmath337 , @xmath159}|x^{\delta , i}_{t , r}|^p\bigr)\leq c_{p,\sa,\sk } |t|\me\bigl(\sup_{r\in[t,0]}|x^{\delta , i}_{t , r}|^p\bigr).\ ] ] from this , we deduce that there exists a time @xmath338 such that for all @xmath339 $ ] , @xmath340 } \me\bigl(\sup_{r\in[t,0]}|x^{\delta , i}_{t , r}(x)|^p\bigr)<+\infty .\ ] ] on the other hand , let @xmath341 satisfy the following sfde : @xmath342 \\[-8pt ] \nonumber & & { } + \int^s_t\nabla_y g_r(x_{t , r},\me^{\sf_{r-}}(\phi _ r(x_{t,\cdot } ) ) ) \\ & & \hspace*{28pt}{}\times \me^{\sf_{r-}}\biggl(\nabla\varphi(x_{t,0})y^i_{t,0 } + \int^0_r\nabla f_r'(x_{t , r'})y^i_{t , r'}\,\dif r'\biggr)\,\dif l_r,\nonumber\end{aligned}\ ] ] which can be solved on @xmath343 $ ] as in theorem [ th1 ] . using the uniform estimate ( [ el2 ] ) and formula ( [ for ] ) ,
it is not hard to deduce that @xmath344}|x^{\delta , i}_{t , r}(x)-y^i_{t , r}(x)|^p\bigr)=0.\ ] ] in particular , @xmath345}\me\bigl(\sup_{r\in [ t,0]}|y^i_{t , r}(x)|^p\bigr)<+\infty.\ ] ] the higher derivatives can be estimated similarly from ( [ nn3 ] ) .
now we can prove the following result , which was originally due to .
[ th5 ] assume that holds for some @xmath287 , and holds for all @xmath144 .
let @xmath346 be the maximal solution of sfde ( [ sde2 ] ) in theorem [ th2 ] , and @xmath26 be defined by @xmath347 then there exists a time @xmath348 such that for each @xmath339 $ ] , @xmath284 has bounded derivatives up to ( @xmath349)-order , and solves the following quasi - linear partial integro - differential equation : @xmath350\,\dif s\qquad \forall(t , x)\in[t_*,0]\times\mr^d,\ ] ] where @xmath351 and @xmath352\nu(\dif z).\ ] ] we follow the argument of friedman @xcite . by proposition
[ pro2 ] , for @xmath353 , we have @xmath354 + \me\biggl[\me\biggl(\int^0_t f_r(x_{t , r}(y))\,\dif r\biggr)\bigg|_{y= x_{t - h , t}(x)}\biggr]\\ & & { } + \me\biggl(\int^t_{t - h } f_r(x_{t - h , r}(x))\,\dif r\biggr)\\ & = & \me u_t(x_{t - h , t}(x))+\me\biggl(\int^t_{t - h } f_r(x_{t - h , r}(x))\,\dif r\biggr).\end{aligned}\ ] ] by lemma [ th4 ] , it is easy to see that there exists a time @xmath355 such that for each @xmath339 $ ] , @xmath26 has bounded derivatives up to @xmath356-order .
thus , we can invoke lemma [ le2 ] to derive that @xmath357 on the other hand , from the above proof , it is also easy to see that for fixed @xmath28 , @xmath358 is lipschitz continuous . hence , @xmath359\,\dif s.\ ] ] the proof is thus complete . in this subsection
we assume that ( * * h**@xmath273 ) holds for some @xmath360 , and @xmath361 and @xmath102 are uniformly bounded and continuous functions .
consider the following sfde : @xmath258 } \int_{|z|<1}\cg_{t , r}(x)\cdot z\tilde n(\dif r,\dif z ) + \int_{(t , s ] } \int_{|z|\geq1}z n(\dif r,\dif z)\\ & & { } + \int_{(t , s]}\cg_{t , r}(x)\cdot b\,\dif r+\int_{(t , s]}\cg _ { t , r}(x)\,\dif w^a_r,\end{aligned}\ ] ] where @xmath362 is defined by ( [ ep0 ] ) . in this case , lemmas [ le2 ] and [ th4 ] still hold .
we just want to mention that ( [ el4 ] ) should be replaced by @xmath363 and ( [ eu2 ] ) becomes @xmath364 } \int_{|z|\leq1}\cg^{\delta , i}_{t , r}\cdot z\tilde n(\dif r,\dif z ) + \int_{(t , s]}\cg^{\delta , i}_{t , r}\cdot b\,\dif r+\int_{(t , s]}\cg ^{\delta , i}_{t , r}\,\dif w^a_r.\ ] ] thus , the following result can be proved along the same lines as in theorem [ th5 ] .
we omit the details .
[ th8 ] assume that holds for some @xmath287 , and @xmath361 and @xmath102 are uniformly bounded and continuous functions .
let @xmath365 be the short time solution of sfde ( [ sde2 ] ) in theorem [ th22 ] , and @xmath26 be defined by ( [ ee ] ) .
then there exists a time @xmath348 such that for each @xmath339 $ ] , @xmath284 has bounded derivatives up to ( @xmath349)-order , and solves the following quasi - linear partial integro - differential equation : @xmath350\,\dif s\qquad \forall(t , x)\in[t_*,0]\times\mr^d,\ ] ] where @xmath366 and @xmath367\\ & & \hspace*{23pt}{}\times \nu(\dif z)\\ & & { } + \int_{|z|\geq1}[u_t(x+z)-u_t(x)]\nu(\dif z).\end{aligned}\ ] ]
in this section we consider the following semi - linear partial integro - differential equation : @xmath368 where @xmath36 is the generator of the lvy process @xmath75 given by ( [ ge ] ) , and @xmath369 \\[-8pt ] \nonumber \varphi&\in&\mw^{1,\infty}(\mr^d;\mr^k).\end{aligned}\ ] ] here and below , @xmath370 denotes the space of bounded and lipschitz continuous functions , @xmath371 or @xmath372 denotes the space of uniformly or locally bounded measurable functions .
let us first give the following definition about the maximal weak solution for equation ( [ ee1 ] ) .
[ ee4 ] for @xmath373 , we call @xmath374 ; \mw^{1,\infty } ( \mr ^d;\mr^k))$ ] a maximal weak solution of equation ( [ ee1 ] ) if @xmath375 and for all @xmath376 and @xmath120 $ ] , @xmath377 where @xmath378 , and @xmath379 is the adjoint operator of @xmath36 and given by @xmath380\nu(\dif z).\ ] ] the main aim of this section is to prove the following existence and uniqueness of a maximal weak solution as well as the global solution for equation ( [ ee1 ] ) .
[ main ] ( locally maximal weak solution ) under ( [ ee3 ] ) , there exists a unique maximal weak solution @xmath26 for equation ( [ ee1 ] ) in the sense of definition [ ee4 ]
. moreover , let @xmath44 be the maximal existence time , then for any @xmath120 $ ] , @xmath381}\|f_s\| _ \infty .\ ] ] ( nonnegative solution ) if for some @xmath276 , the components @xmath382 and @xmath383 are nonnegative , then the corresponding component @xmath384 of weak solution in are also nonnegative .
( global solution ) let @xmath77 be the lvy symbol defined in ( [ lp9 ] ) with @xmath385 . if for some @xmath386 , @xmath387 where @xmath388 means that for some @xmath389 , @xmath390 , then the maximal existence time @xmath44 in equals to @xmath391 . in the case that @xmath392 and @xmath393 is strictly positive , then @xmath44 also equals to @xmath391 .
since we have estimate ( [ as ] ) , it is easy to see that the assumption on @xmath33 in ( [ ee3 ] ) can be replaced by @xmath394 where @xmath395 . for proving this theorem ,
let us begin with studying : in this subsection , we firstly study the existence and uniqueness of weak solutions for the following linear pide : @xmath396 where @xmath397 , @xmath398 , @xmath242 and @xmath241 are bounded measurable functions .
let us start with the following case of smooth coefficients , which is the classical feynman
kac formula . here , the main point is to prove the uniqueness .
[ ee7 ] assume that @xmath399 where @xmath400 denotes the space of bounded smooth functions with bounded derivatives of all orders .
let @xmath401 solve the following sde : @xmath402 and @xmath403 solve the following ode : @xmath404 define @xmath405+\me \biggl[\int^0_t z_{t , r}(x)f_r(x_{t , r}(x))\,\dif r\biggr].\ ] ] then @xmath406 uniquely solves the following linear pide : @xmath407\,\dif s \nonumber \\[-8pt ] \\[-8pt ] \eqntext { \forall(t , x)\in\mr_-\times\mr^d.}\end{aligned}\ ] ] by smoothing the time variable and then taking limits , as in section [ sec3 ] , by careful calculations , one can find that @xmath19 defined by ( [ ee5 ] ) belongs to @xmath408 and satisfies ( [ ee111 ] ) ; see @xcite , page 148 , theorem 5.3 .
we now prove the uniqueness by the duality argument .
let @xmath409 solve the following sde : @xmath410 and @xmath411 solve the following ode : @xmath412\cdot\hat z_{t , r}(x)\,\dif x.\ ] ] fix @xmath117 and @xmath376 define @xmath413.\ ] ] as above , one can check that @xmath414\,\dif s.\ ] ] let @xmath406 satisfy equation ( [ ee111 ] ) with @xmath415 . then by the integration by parts formula , we have for almost all @xmath121 $ ] , @xmath416 from this , we get @xmath417 which leads to @xmath418 by the arbitrariness of @xmath419 . for @xmath420
, we introduce a family of mollifiers in @xmath421 .
let @xmath422 $ ] be a smooth function satisfying that @xmath423 we shall call @xmath424 a family of mollifiers in @xmath421 .
next , we relax the regularity assumptions on @xmath425 and @xmath224 , and prove the following : [ th6 ] assume that @xmath426 let @xmath26 be defined as in ( [ ee5 ] )
. then @xmath427 is a unique weak solution of equation ( [ ee11 ] ) in the sense of definition [ ee4 ] .
we only prove the uniqueness . as for the existence
, it follows by smoothing the coefficients and then taking limits as done in theorem [ th7 ] below .
suppose that @xmath428 is a weak solution of equation ( [ ee11 ] ) with @xmath415 in the sense of definition [ ee4 ] .
we want to prove that @xmath429 .
let @xmath430 be a family of mollifiers in @xmath3 .
define @xmath431 taking @xmath432 in ( [ ep9 ] ) , one finds that @xmath433 satisfies @xmath434\,\dif s,\ ] ] where @xmath435 by the property of convolutions , we have @xmath436 and for fixed @xmath51 and lebesgue almost all @xmath28 , @xmath437 let @xmath438 solve the following sde : @xmath439 and @xmath440 solve the following ode : @xmath441 by theorem [ ee7 ] , @xmath433 can be uniquely represented by @xmath442.\ ] ] for completing the proof , it suffices to prove that for each @xmath443 , @xmath444 since by ( [ nn1 ] ) , @xmath440 is uniformly bounded with respect to @xmath445 and @xmath115 $ ] , we need only to show that for any nonnegative @xmath446 , @xmath447\to0 , \qquad \eps\to0.\ ] ] for any @xmath448 , by ( [ uu0 ] ) and the change of variables , we have @xmath449 \nonumber \\[-8pt ] \\[-8pt ] \nonumber & \leq&\me\biggl[\int^0_t \int_{b_r } , -1}_{t , s}(x))\,\dif x\,\dif s\biggr ] \\ & & { } + c_t\int^0_t \int_{\mr^d}p\{|x^\eps _ { t , s}(x)|>r\ } \psi(x)\,\dif x\,\dif s,\nonumber\end{aligned}\ ] ] where @xmath450 denotes the inverse of @xmath451 . from equation ( [ nn2 ] ) , it is by now standard to prove that ( e.g. , see kunita @xcite , lemma 4.3.1 ) @xmath452 which then yields @xmath453}|\det ( \nabla x^{\eps,-1}_{t , s}(x))|<+\infty.\ ] ] thus , for fixed @xmath448 , the first term in ( [ uu1 ] ) is less than @xmath454 moreover , by equation ( [ nn2 ] ) , we also have @xmath455 the proof is complete by first letting @xmath456 and then @xmath457 in ( [ uu1 ] ) . as an easy corollary of this theorem
, we first establish the uniqueness for equation ( [ ee1 ] ) . under ( [ ee3 ] ) , there exists , at most , one weak solution for equation ( [ ee1 ] ) .
let @xmath458 ; \mw^{1,\infty}(\mr^d;\mr^k ) ) , i=1,2 $ ] be two weak solutions of equation ( [ ee1 ] ) in the sense of definition [ ee4 ] .
define @xmath459 .
then @xmath26 satisfies that for all @xmath376 , @xmath460 where @xmath461 by ( [ ee3 ] ) , it is easy to verify that @xmath462;\mw^{1,\infty } ( \mr ^d;\mr^d))\ ] ] and @xmath463;\cb(\mr^d;\mr^k)).\ ] ] thus , by theorem [ th6 ] , we conclude that @xmath464 .
consider the following sfde : @xmath466 \\[-8pt ] \nonumber & & { } + \int^s_t\,\dif l_r,\hspace*{-35pt}\end{aligned}\ ] ] where @xmath467 , @xmath468 and @xmath241 are bounded measurable functions .
we need the following continuous dependence of the solutions with respect to the coefficients .
[ pro3 ] suppose that @xmath469 are two groups of bounded measurable functions , and for some @xmath470 and all @xmath471 , @xmath127 and @xmath472 , @xmath473 then there exists a time @xmath117 depending only on @xmath474 such that for all @xmath121 $ ] and @xmath475 , @xmath476}\me\bigl|x^{(1)}_{t , s}(x)-x^{(2)}_{t , s}(y)\bigr|\nonumber\\ & & \qquad\leq2|x - y| + 2\int^0_t\bigl\|g^{(1)}_r - g^{(2)}_r\bigr\|_\infty\,\dif r \\ & & \qquad\quad { } + 2k|t|\biggl(\bigl\|\varphi^{(1)}-\varphi^{(2)}\bigr\|_\infty+\int^0_t \bigl\| f^{(1)}_r - f^{(2)}_r\bigr\|_\infty\,\dif r\biggr),\nonumber\end{aligned}\ ] ] where @xmath477 is the solution of ( [ sfee ] ) corresponding to @xmath478 .
set @xmath479 by ( [ sfee ] ) and the assumptions , we have @xmath480}\me|z_{t , r}|.\end{aligned}\ ] ] from this , we immediately conclude the proof .
[ th7 ] assume that @xmath481 are bounded measurable functions and satisfy for some @xmath470 and all @xmath471 , @xmath127 and @xmath472 , @xmath482 then there exists a time @xmath483 such that @xmath484 is a unique weak solution of equation ( [ ee1 ] ) on @xmath119 $ ] in the sense of definition [ ee4 ] .
let ( @xmath485 ) be the smooth approximation of ( @xmath486 ) defined by @xmath487 where @xmath488 [ resp . , @xmath489 and @xmath490 are the mollifiers in @xmath491 ( resp . , @xmath492 and @xmath3 ) .
it is clear that @xmath493 by theorem [ th3 ] and proposition [ pro3 ] , there exists a time @xmath494 such that for all @xmath495 and @xmath28 , @xmath496 where @xmath497 ( resp . , @xmath142 )
is the solution family of sfde ( [ sfee ] ) corresponding to the coefficients ( @xmath498 ) [ resp . , ( @xmath486 ) ] .
using this limit , and by the dominated convergence theorem , it is easy to verify that for each @xmath499\times\mr^d$ ] , @xmath500 where @xmath433 is defined through @xmath501 and @xmath502 as in ( [ eu ] ) .
moreover , by proposition [ pro3 ] , we also have @xmath503}\|\nabla u^\eps_t\|_\infty + \sup_{t\in[t,0]}\|\nabla u_t\|_\infty\leq c_{t , k}<+\infty.\ ] ] on the other hand , thanks to ( [ eu1 ] ) , by theorem [ th8 ] , there exists another time @xmath504 independent of @xmath505 such that @xmath506 in particular , for all @xmath507 and all @xmath508 $ ] , @xmath509 \\[-8pt ] \nonumber & & { } + \int ^0_t\langle f^\eps_r,\psi\rangle\,\dif r.\end{aligned}\ ] ] we want to take limits for both sides of the above identity by ( [ ll1 ] ) .
the key point is to prove @xmath510 which will be obtained by proving the following two limits : @xmath511 the first limit is clear by ( [ ll1 ] ) , ( [ eu1 ] ) and the dominated convergence theorem
. the second limit follows by ( [ ll1 ] ) , ( [ eu1 ] ) and the integration by parts formula .
now we are in a position to give : we divide the proof into three steps .
( _ step _ 1 ) . for @xmath512 , define @xmath513 and @xmath514 in this step , we prove the following claim : [ cl ] for given @xmath515 , there exists a time @xmath516 such that for any bounded measurable function @xmath517 satisfying @xmath518}\|\nabla h_t\|_\infty\leq u$ ] , it holds that @xmath519}\| f^h_s\| _ \infty,\ ] ] and @xmath520}\|\nabla u^h_t\|_\infty\leq u,\ ] ] where @xmath521 is defined by ( [ eu ] ) in terms of @xmath522 and @xmath523 , and @xmath524 is the unique solution family of sfde ( [ sfee ] ) corresponding to ( @xmath525 ) .
proof of the by proposition [ pro3 ] , there exists a time @xmath526 such that for all @xmath475 , @xmath527 using this and by the definition of @xmath521 [ see ( [ eu ] ) ] , we have @xmath528
so , @xmath529}\|\nabla u^h_s\|_\infty&\leq&2\|\nabla\varphi\| _
\infty+ 2|t|\sup_{s\in[t,0]}(\|\nabla_x f_s\|_\infty+\|\nabla_u f_s\| _ \infty u)\\ & \leq&2\|\nabla\varphi\|_\infty+2|t|\sk(u+1).\end{aligned}\ ] ] since @xmath515 , choosing @xmath530 , we obtain ( [ epp1 ] ) .
estimate ( [ epp2 ] ) follows from definition ( [ eu ] ) .
( _ step _ 2 ) .
set @xmath531 .
we construct the following iteration approximation sequence : for @xmath148 , @xmath532 by the above claim , there exists a time @xmath533 such that for all @xmath148 , @xmath534}\|f_s\| _ \infty , \qquad \sup_{t\in[t_1,0]}\|\nabla u^n_t\|_\infty\leq4\|\nabla\varphi\| _ \infty .\ ] ] hence , @xmath535 thus , by the definition of @xmath536 [ see ( [ eu ] ) ] and proposition [ pro3 ] again , there exists another time @xmath537 such that for all @xmath538 and @xmath141 , @xmath539 where @xmath42 is independent of @xmath540 . by gronwall s inequality , we obtain that @xmath541}\|u^n_t - u^m_t\|_\infty=0.\ ] ] hence , there exists a @xmath542\times\mr^d;\mr^k)$ ] such that @xmath543}\|u^n_t - u_t\|_\infty=0,\ ] ] and by ( [ epo1 ] ) , @xmath544}\|f_s\| _ \infty ,
\qquad \sup_{t\in[t,0]}\|\nabla u_t\|_\infty\leq4\|\nabla\varphi\| _ \infty.\ ] ] on the other hand , by theorem [ th7 ] , @xmath536 satisfies that for all @xmath376 , @xmath545 thus , one can take limits as in theorem [ th7 ] to obtain the existence of a short time weak solution for equation ( [ ee1 ] ) .
moreover , ( ii ) follows from ( [ eu ] ) .
the existence of a maximal weak solution can be obtained as in the proof of theorem [ th2 ] by shifting the time and the induction .
thus , we conclude the proof of ( i ) . as for (
ii ) , it follows by ( [ er1 ] ) and the definition of @xmath536 .
( _ step _ 3 ) .
let @xmath374;\mw^{1,\infty}(\mr^d;\mr^k))$ ] be a maximal weak solution of equation ( [ ee1 ] ) .
define for @xmath546\times\mr^d$ ] , @xmath547 then it is clear that @xmath548;\mw^{1,\infty}(\mr^d;\mr^d)),\qquad f\in\cb_{\mathrm{loc}}((t,0];\mw^{1,\infty}(\mr^d;\mr^k)).\ ] ] for @xmath120 $ ] , let @xmath401 solve the following sde : @xmath549.\ ] ] define @xmath550 by theorem [ th6 ] , we have @xmath551\times\mr^d.\ ] ] suppose now that @xmath270 . for completing the proof , by ( [ max ] )
it is enough to show that @xmath552 it immediately follows from ( [ ei1 ] ) and the following claim proved in @xcite , theorem 4.5 , which is stated in a slight variant . under ( [ eu3 ] ) or @xmath393 nondegenerate , for any bounded continuous function @xmath224 and @xmath210 , @xmath553 where @xmath554 only depends on @xmath555 and the bound of @xmath556 .
deep thanks go to the referee for his / her very careful reading the manuscript and very useful suggestions . | in this article we study a class of stochastic functional differential equations driven by lvy processes ( in particular , @xmath0-stable processes ) , and obtain the existence and uniqueness of markov solutions in small time intervals .
this corresponds to the local solvability to a class of quasi - linear partial integro - differential equations .
moreover , in the constant diffusion coefficient case , without any assumptions on the lvy generator , we also show the existence of a unique maximal weak solution for a class of semi - linear partial integro - differential equation systems under bounded lipschitz assumptions on the coefficients .
meanwhile , in the nondegenerate case ( corresponding to @xmath1 with @xmath2 $ ] ) , based upon some gradient estimates , the existence of global solutions is established too .
in particular , this provides a probabilistic treatment for the nonlinear partial integro - differential equations , such as the multi - dimensional fractal burgers equations and the fractal scalar conservation law equations . |
_ asca _ observations have proved to be instrumental in our understanding of narrow - line seyfert 1 galaxies ( nls1s ) .
one of the most compelling first indications that nls1s may be characterized by a high accretion rate came from the _ asca _ spectrum of re 1034 + 39 , reported by pounds , done & osborne ( 1995 ) .
this spectrum revealed a strong soft excess component and a steep hard x - ray power law that seemed to be similar to the spectra of black hole candidates in the high state .
while it is now quite clear that not all nls1s have such spectra and indeed a range of strengths of the soft excess are seen ( e.g. leighly 1999b ) , and that perhaps comparison with the _ very _ high state seen in some galactic x - ray emitting objects may be more appropriate , the result has been extremely important in the development of our understanding of these objects .
the results from the _ asca _ observations of nls1s have now been reported in several places ( e.g. leighly 1999ab and references therein , see also vaughn et al . these proceedings ) .
therefore , only part of this review will be devoted to a few results from the variability analysis of nls1s and a very brief discussion of models and implications . in the second half ,
i present a preview of our work on the _ hst _ spectra from nls1s ( leighly & halpern 2000 ) .
the x - ray variability in nls1 _ asca _ observations is discussed in detail in leighly 1999a .
because of the gaps in the light curves , the _ excess variance _ , also known as the fractional amplitude of variability , was used to quantify the variability .
the results are shown in fig . 1 ( left ) .
this figure shows that for a given x - ray luminosity , the amplitude of variability is consistently larger for nls1s than for seyfert 1 galaxies with broad optical lines ( bls1s ) .
the uniform sampling and other properties of the _ asca _ data mean that the excess variance should be proportional to the inverse of a time scale of variability @xmath2 as @xmath3 , where @xmath4 is the slope of the variability power spectrum ( see also lawrence & papadakis 1993 ) . thus , the simplest interpretation of fig .
1 is that , if the hard x - ray luminosity is characteristic of the absolute mass accretion rate ( i.e. the efficiency of conversion of accretion energy to radiation is the same in all objects ) , then nls1s have a shorter variability time scale than do seyfert 1s with broad optical lines ( bls1s ) .
this would be equivalent to compressing a bls1 light curve into an interval approximately 10 times shorter in nls1s .
the variability time scale may be characteristic of the black hole mass , simply from light - crossing time - scale arguments .
thus , the simplest explanation of this result is that nls1s have a smaller black hole mass than bls1s , and since they are no fainter than bls1s , they must be accreting at a relatively larger accretion rate .
closer examination indicates that the situation may not be as simple as the argument above would imply .
specifically , fig .
1 shows a large spread in the excess variance values for nls1s , much larger than the measurement error , but also much larger than the model - dependent systematic error in @xmath5(excess variance ) of @xmath60.310.47 arising from the weakly nonstationary nature of a @xmath7 power spectrum ( see leighly 1999a for details ) .
it is possible that the large spread is a consequence of the fact that the nature of the variability may not be homogeneous among nls1s .
leighly 1999b report a correlation between the strength of the soft excess in the _ asca _ spectrum and the amplitude of the variability , and suggests that strong soft excess objects are characterized by flaring , high amplitude variability . in fig
. 1 ( right ) , i separate the 6 objects identified as having strong soft excesses in leighly 1999b and plot the excess variance as a function of 0.52.0 kev luminosity . when these six objects are excluded , the slope of the regression at @xmath8 is no longer biased by the scatter .
the slope is still somewhat flatter than that expected if the time scale is proportional to the luminosity assuming that the slope of the variability power spectrum is @xmath9 ( dashed lines ) .
what is the origin of the enhanced variance observed in nls1s ? as stated above , it may be simply a consequence of a smaller emission region characteristic of a smaller black hole .
however , the light curves in some strong soft excess nls1s are characterized by very high amplitude flares , and they could not be simply time - compressed versions of bls1 light curves .
a few other emission mechanisms have been suggested that instead would be equivalent to stretching the amplitude of flares in bls1 lightcurves to produce the enhanced excess variance observed in nls1s . naturally , because of the scale invariant nature of the variability power spectrum ( at least in the range of frequencies that the _ asca _ data probe ) these general scenarios can not be distinguished using the power spectrum or the excess variance . * noting that the very large amplitude variability on short time scales implies a very high efficiency of conversion of accretion energy to radiation in the nls1 phl 1092 , it has been proposed that beaming plays an important role in amplifying flares ( brandt et al.1999 ) .
a specific scenario has been proposed : the emission regions are found on the very inner edge of the accretion disk , and the accretion disk is inclined at a very high angle with respect to the viewer so that doppler effects cause amplification of flares ( boller et al . 1997 ) .
the results presented in the next section of this contribution cast significant doubt on an edge - on orientation . * it has recently been suggested that if nls1s are accreting at a higher rate then the magnetic field energy , assumed to be in equipartion with gravitational potential energy , should be proportionally larger ( mineshige et al .
large amplitude flares may result from reconnection of this more powerful magnetic field . *
it has been suggested that occultations by large optically - thick clouds may produce high - amplitude x - ray variability ( brandt et al.these proceedings ) .
an observed steep @xmath10 in some objects seems to support this idea ; however , there is no physical reason that luminous nls1s should not have intrinsically steep @xmath10 , for example , if the x - ray emitting corona is weak or not present .
what are the implications of the high amplitude flaring ?
it has been previously suggested that high amplitude flaring is evidence that the variability is nonlinear ( green , mchardy & done 1999 ; boller et al.1997 ) .
this conclusion is based on the assumption that the agn variability is gaussian , since in that case a very high variance compared with the mean can only be produced if the variability is nonlinear .
in fact , non - gaussian variability makes more sense for agns because our usual physical picture supposes that the light curve should be built up from the superposition of flares , and flares are inherently non - gaussian .
linear , non - gaussian variability can easily produce high values of excess variance .
non - gaussian variability can be detected using a parameter related to the skew of the flux distribution , and evidence for non - gaussianity was found in several of the nls1s with the highest amplitude of variability .
demonstration that a light curve is nonlinear is a much harder problem ; it is a very important one , however , because of the potentially strong constraints on emission processes and geometry .
see leighly 1999a for a detailed discussion .
we find some evidence for nonlinear variability in the broad - line radio galaxy 3c 390.3 ( leighly & obrien 1997 ) in that quiescent periods occur before and after large flares , and this behavior can not be reproduced by a linear non - gaussian model .
evidence for nonlinear variability has been recently reported in cyg x-1 ( timmer et al .
in 1997 , we reported the detection of an absorption feature near 1 kev in the _ asca _ spectra from three nls1s , iras 13224@xmath13809 , 1h 0707@xmath1495 , and pg 1404 + 226 ( leighly et al . 1997 ) .
absorption by ionized oxygen is common in the x - ray spectra from seyfert 1 galaxies ; however , if we are to interpret the 1 kev features in this way , the absorbing material must have highly relativistic velocities ( 0.20.6@xmath11 , depending on whether the features are interpreted as absorption lines or edges ) .
acceleration of ionized gas to these high velocities should be difficult ; however , we noted that nls1s exhibit suggestive similarities to a subclass of broad - absorption line quasars ( balqsos ) , objects characterized by the signature of absorption by outflowing gas in the uv : both types of objects exhibit strong or extreme and weak [ ] emission , they often have red continua and strong infrared emission , and they are predominately radio - quiet ( see leighly et al . 1997 for details ) .
an alternative explanation as absorption by highly ionized neon and iron has been presented by nicastro et al .
( 1999 ) . with luck
, the origin of these features will be resolved by our upcoming _
chandra _ observation of 1h 0707@xmath1495 .
absorption in x - rays is often accompanied by absorption in the uv . to test our hypothesis that the x - ray absorption features are produced by relativistically outflowing gas , we applied for and were awarded _ hst _
stis uv spectroscopic observations of iras 13224@xmath13809 and 1h 0707@xmath1495 ( leighly & halpern 2000 ) and the results of the observations were presented for the first time at this meeting .
the broad - band continuum spectra , corrected for galactic reddening , are presented in fig . 2 along with the average qso spectrum compiled by paul francis .
this figure shows that these nls1s have continua as blue as the average qso3809 has been previously reported as having a red spectrum ( mas - hesse et al .
we find that our optical spectrum obtained from a 1 hour exposure on the ctio 4 meter telescope is indeed rather red and does not join smoothly to the uv spectrum .
we infer , from a difference in the spatial profile of the lines and continuum , that there is a strong , red galaxy component in the spectrum .
in contrast , our 1h 0707@xmath1495 optical spectrum is quite blue and joins smoothly onto the uv spectrum . ] .
there is also no evidence for resonance - line absorption intrinsic to the object ; all of the narrow absorption lines in the spectra originate in our galaxy .
the emission line profiles are the most interesting feature of these spectra .
the high - ionization lines , including ly@xmath4@xmath12 , @xmath13 , @xmath14 and @xmath15 have a different profile than the low - ionization lines , including @xmath16 and h@xmath17@xmath18 .
the difference is illustrated in fig .
3 which overlays the rescaled high - ionization line and the low - ionization line profiles as a function of velocity2796 , @xmath192804 is consistent with that of h@xmath17 . ] .
as is commonly found in agn ( e.g. marziani et al .
1997 ) , the average quasar high - ionization lines are broader and blueshifted relative to the low - ionization lines .
the same trend is true in nls1s , except the difference is more profound .
the fwhm is @xmath20 that of h@xmath17 , but it is significant that all of the extra width is on the blue side , whereas the red side lines up well with the low - ionization lines .
similar line profiles were reported from the nls1 i zw 1 ( laor et al .
1997 ) .
we also find evidence for emission from gas with very high densities . in fig .
4 we plot the ] ] line region of our spectra , as well as archived spectra from several other nls1s .
we find a very high ] to ] ratio in our spectra from iras 13224@xmath13809 and 1h 0707@xmath1495 .
the ions responsible for these lines are found under the same physical conditions ; however , ] has a higher critical density than ] and thus the ratio becomes large when the ] emission has saturated .
a correlation between this ratio and the width of h@xmath17 has been previously reported by wills et al .
1999 ( see also wills , these proceedings ) .
we note that these emission lines are relatively narrow and symmetric about their rest wavelengths and thus their profiles are more similar to those of the low - ionization lines .
4 compares line profiles from 6 nls1s .
three of the objects ( i zw 1 , iras 13224@xmath13809 and 1h 0707@xmath1495 ) display blueshifted high - ionization lines and high ] to ] ratios .
rx j0134@xmath142 also appears to have a blueshifted line .
however , the spectra from two lower luminosity nls1s , akn 564 and re 1034 + 39 , lack both of these attributes .
we suspect that this reflects a dependence of the line properties on luminosity .
we propose that the characteristic spectra we observe in the higher luminosity nls1s can be naturally explained by a disk - wind model , as shown schematically in fig .
5 . it has long been known that it is not possible to produce all broad - line emission in gas characterized by one set of physical properties .
several multicomponent models , including disk - wind models , have been proposed ; unfortunately , space limitations prohibit a through review here . what is new here is that our observations provide evidence that a disk - wind model is strongly favored over all other models .
the fact that the red side of the high- and low - ionization lines line up suggests that the low - ionization lines and low - velocity cores of the high - ionization lines are produced in the same relatively low - velocity material ( i.e. the accretion disk ) while the high - velocity blue wings of the lines are produced in a wind .
the disk is optically thick , and therefore , we see only emission from the wind accelerated toward us .
we are currently constructing a simple disk - wind model to test our assertion and fig .
5 shows a schematic of the assumed geometry .
we are following murray & chiang ( 1998 ) in general : we use a @xmath17 velocity law , often applied in star and cv winds , and we are using _ cloudy _ to estimate the ionization structure .
however , since we are interested in the emission produced in the wind in particular , we treat the radiative transfer numerically using the sobolev approximation . a complementary and more sophisticated model being constructed by proga , stone and kallman ( 2000 ; hereafter psk ) aims to determine the detailed dynamics of the wind .
a definitive model has not yet been constructed ; however , it is clear that there are potentially complex relationships between the radiative transfer , dynamics and ionization , and the results will quite possibly not conform to naive intuition when everything has been taken into account .
examples of two complications : * the wind is quite likely to be optically thick to scattering of the resonance lines , and thus radiative transfer will be strongly enhanced along directions in which the velocity gradient is high .
this effect can strongly alter the observed line profiles for the high - ionization lines emerging from the wind , and prompts the use of the sobolev approximation .
the recombination lines are immune to this effect , however . *
the acceleration of the wind is produced by resonance - line driving and thus the velocities attained depend intimately on the ionization structure of the gas .
specifically , there must be a mechanism for shielding the wind gas from the photoionizing x - ray source .
this problem seems to have a natural solution : psk find that there can be adequate `` failed '' wind material falling back onto the object to do this .
our preliminary results and also the results presented by psk are promising enough to conjecture that many of the features of the _ hst _ nls1 spectra will be explainable using a disk - wind system .
for example : * a disk - wind model coupled with a high accretion rate may naturally explain nls1 spectra .
psk find that a primary condition on formation of the wind is that the wind - driving ( uv ) flux should be large compared with the wind - photoionizing ( x - ray ) flux .
a high accretion rate predicts a spectrum strongly peaked in the extreme uv and therefore the uv to x - ray ratio will be large enough that a strong wind can be driven .
if the accretion rate is high , then densities in the disk will be high , resulting in a large ] to ] ratio .
more material will be available to be blown from the nucleus , and since the black hole mass is relatively smaller , the wind is more likely to reach escape velocity .
the wind may partially shield the accretion disk from the intense radiation field , thus permitting excitation but not causing ionization of the fe@xmath21 ion , so that strong optical emission may be produced .
this could explain the observed correlation between high - ionization blue asymmetry and optical ( e.g. marziani et al .
* the large uv to x - ray ratio requirement found by psk may explain why lower luminosity nls1s do not show blue wings on their high - ionization lines .
quasars are known to have typically relatively larger uv to x - ray ratios than seyferts ( wilkes et al .
1994 ) , and thus gas in seyferts may become too highly ionized to be accelerated to high velocities . *
the observed line profiles will be affected by the observer s viewing angle , and radiative transfer effects , as noted above , potentially affect the profiles as well in a non - obvious way .
we interpret the fact that the red sides of the high- and low - ionizations lines line up as evidence that the low - ionization lines and the low - velocity core of the high - ionization lines are produced in the disk .
however , for there to be only blue - side wind emission , either the observer s viewing angle must be nearly face on , or the angle of the wind stream lines with respect to the disk must be large .
it is quite possible that in reality the wind stream line angle and the radius at which the wind is launched are not independent of the accretion rate , and thus results from a self - consistent model would be potentially very interesting .
we have described how the particular uv spectra from higher luminosity nls1s strongly suggests that the emission lines are produced in a disk - wind system .
further , we conjecture that , through a potentially complicated dependence of model parameters such as the amount of material and velocity and angle of stream lines on the physical parameters such as the black hole mass and accretion rate , it may be possible to explain eigenvector 1 self - consistently .
there are a few more important issues : * broad emission lines in agn are often assumed to be produced in clouds that have virialized bulk motions ( e.g. peterson & wandel 1999 ) .
in contrast , the strong winds that we propose are responsible for the blue wings in our spectra and thus for a significant fraction of the high - ionization line emission in luminous nls1s will not have virialized bulk motion . *
it has been proposed that nls1s bear some similarity to high redshift quasars ( mathur 2000 ) : in particular , very high redshift quasars ( @xmath22 ) appear to have rather narrow lines compared with local quasars ( shields et al .
the spectra presented here show that this conjecture may have serious weaknesses . at high redshift
, only the rest - frame uv high - ionization lines are observed in the optical spectrum .
we have discovered that the high - ionization lines in higher - luminosity nls1s are typically broad ; furthermore , wills et al .
( these proceedings ) find no correlation between the high - ionization and h@xmath17 line widths in the pg - quasar subsample .
these results imply that the observed optical spectra of high - redshift nls1s will show _ broad _ emission lines , and therefore , such objects may be difficult to identify . finally , if there is a wind , it is possible that some of the emission is a result of excitation of high velocity n@xmath23 ions by absorption of ly@xmath4 photons , and abundance enhancements may be less necessary ( however , see also hamann & korista 1996 ) .
* will the wind be homogeneous ?
resonance - line driven winds are a characteristic of wolf - rayet stars and there is very strong evidence for density enhancements in these winds .
furthermore , psk find density inhomogeneities in their simulations .
clumping may be necessary to match the intensity of observed emission lines .
if present , clumping could produce interesting time - dependent and ionization - dependent behavior .
* acknowledgements : * i would like to thank the wilhelm and else heraeus foundation for travel support , jules halpern for helpful discussions , and daniel proga for an early look at his submitted paper .
i gratefully acknowledge support through nag5 - 7971 ( nasa ltsa ) .
shields , j. c. , hamann , f. , foltz , c. b. , & chaffee , f. h. , 1997 , proc .
`` emission lines in active galaxies : new methods and techniques '' , ed . b. m. peterson , f .- z .
cheng & a. s. wilson ( aps : san francisco ) , p. 118 | the study of nls1s using _ asca _ has many advantages .
a representative sample can be studied ; to date , observations of more than 30 nls1s have been made .
_ asca _ observations are conducted contiguously , so their x - ray variability properties can be studied systematically .
_ asca _ detectors have a broad band pass and moderate energy resolution , properties which allow their complex x - ray spectrum to be deconvolved .
_ asca _ observations have revealed that a soft excess extending up to 1 kev is frequently found in the x - ray spectra from nls1s .
the hard x - ray photon index has been shown to be systematically steeper than in seyfert 1 galaxies with broad optical lines .
nls1s also exhibit larger amplitude variability than broad - line seyferts of similar hard x - ray luminosity . on occasion , features in the soft x - ray band have been detected that appear to be specific to nls1s ; ionized iron lines have also been observed .
these results have been fundamental for building our current picture of the x - ray emission from nls1s .
many of the properties can be explained if the specific accretion rate is larger in nls1s than in seyfert 1 galaxies with broad optical lines . nls1s
may exemplify an extreme state of an intrinsic physical parameter and therefore their study may lead to an enhanced understanding of agn in general .
this proceedings contribution has two parts .
first i describe and discuss some of the results obtained from a study of the x - ray variability in nls1s using _ asca _ data ( leighly 1999ab ) .
then , i discuss the results from our _ hst _ observations of nls1s that were first presented during this meeting ( leighly & halpern 2000 ) .
# 1#2#1@xmath0#2 galaxies : active galaxies : individual ( iras 13224@xmath13809 , 1h 0707@xmath1495 ) line : profiles ultraviolet : galaxies x - rays : galaxies |
the electronic properties of metals on the border of magnetic phase transitions at low temperatures are often found to exhibit temperature dependences that differ from the predictions of fermi liquid theory .
early attempts to explore such non - fermi liquid behaviour have been based on mean - field treatments of the effects of enhanced magnetic fluctuations , as in the self - consistent renormalization ( scr ) model.@xcite in a recent work the prediction of this model for the temperature dependence of the electrical resistivity @xmath0 was tested in a simple cubic d - metal , ni@xmath8al , at high pressures near to the critical pressure where ferromagnetism is suppressed.@xcite the @xmath9 temperature dependence of the resistivity seen in ni@xmath8al and other related systems , where the magnetic correlation wavevector @xmath10 is small , appears to be largely consistent with the scr model.@xcite in the idealized limit @xmath11 at @xmath12 , the scr model predicts that in 3d the quasiparticle scattering rate @xmath13 varies linearly with the quasiparticle excitation energy , rather than quadratically as in the standard fermi liquid picture .
this form of @xmath13 is similar to that of the marginal fermi liquid model,@xcite which is normally associated with a linear temperature dependence of the resistivity .
however , at the border of ferromagnetism the relevant fluctuations responsible for quasiparticle scattering are of long wavelength and thus are ineffective in reducing the current .
this leads to a transport relaxation rate @xmath14 that differs from @xmath13 and varies not as @xmath15 , but as @xmath9 temperature - composition phase diagram of nis@xmath4se@xmath5.@xcite lower inset shows the temperature dependence of the resistivity of nisse ( @xmath6 ) , which is just on the border of a metallic antiferromagnetic state at low temperatures.@xcite the upper inset is the pyrite crystal structure of nis@xmath1 and nise@xmath1 .
pi and pm stand for paramagnetic insulator and paramagnetic metal , respectively .
afi and afm stand for antiferromagnetic insulator and antiferromagnetic metal , respectively .
sr - afi stands for short range antiferromagnetic insulator.@xcite the temperature - pressure phase diagram of nis@xmath1 is expected to be similar in form to the temperature - composition phase diagram of nis@xmath4se@xmath5 . in this study , the afm boundary of nis@xmath1 is found to be at @xmath2 kbar .
thus , nis@xmath1 at @xmath3 and nisse at ambient pressure may be expected to be similar except for the level of quenched disorder , which is normally higher in nisse than in nis@xmath1 .
the residual resistivity of nisse is typically an order of magnitude higher than in nis@xmath1.,height=226 ] in practice , this form of @xmath14 or of the resistivity @xmath0 can be restricted to a relatively narrow range of temperatures and pressures where @xmath10 is small compared with the characteristic wavevector of thermally excited magnetic fluctuations .
the scr model is not restricted to this limit and can in principle include the effects of @xmath10 and even determine @xmath10 in a self - consistent fashion .
we note that the scr model assumes implicitly that an effective underlying mechanism exists to remove momentum from the electron system .
the validity of this assumption has not been clearly confirmed theoretically , but seems to be consistent with experiment in the cases mentioned above . in its conventional form
the model does not include the possible effects of inhomogeneities or texture that may arise on the border of first order ferromagnetic transitions as in , e.g. , mnsi.@xcite the scr model is also expected to fail on the border of electron localization , as near a mott transition or close to local quantum critical points in heavy electron compounds.@xcite the applicability of the scr model has also been questioned for the case of itinerant - electron antiferromagnetism in general , even well away from the border of a mott transition.@xcite in this paper we present an attempt to test the prediction of the scr model in a metal on the border of metallic antiferromagnetism in 3d for which @xmath16 is predicted to vary as @xmath17 in the idealized limit @xmath11 , @xmath12 , where @xmath18 now stands for the correlation wavevector for the staggered magnetization . the exponent of @xmath19 is the ratio of the spatial dimension @xmath20 and the dynamical exponent @xmath21 .
this may be contrasted with the corresponding exponent of @xmath22 , which is the ratio of @xmath23 and @xmath24 , for the border of metallic ferromagnetism .
( the 2 in @xmath25 arises from the effect of small - angle scattering that is absent in the case of the staggered magnetization . )
this simple model for the scattering from antiferromagnetic fluctuations assumes that the scattering rate can be averaged over the fermi surface . within the scr model
this procedure would seem to require the presence of a sufficient level of quenched disorder assumed to have only the simple consequence of inhibiting the short circuiting caused by the carrier on the cold spots of the fermi surface , i.e. , regions far from the hot spots connected by the antiferromagnetic ordering wavevector and thus strongly affected by spin - fluctuation scattering.@xcite the effect of quenched disorder on the temperature dependence of @xmath16 in the scr model shows up particularly clearly in the temperature - dependent resistivity exponent defined as @xmath26 , where @xmath27 and @xmath28 is the residual resistivity , i.e. , the resistivity extrapolated to @xmath29 k. the resistivity exponent @xmath30 may be described in terms of the fermi wavevector @xmath31 , the elastic mean free path @xmath32 of charge carriers and the reduced temperature @xmath33 , where @xmath34 is a characteristic spin fluctuation temperature.@xcitewith increasing @xmath35 , @xmath30 drops from @xmath19 towards unity around @xmath36 , back towards a value of the order of @xmath19 around @xmath37 and then towards zero for @xmath38.@xcite this undulating behaviour of @xmath30 is a dramatic prediction of the model for relatively pure samples that could in principle be tested by studying a series of samples of the same material with different levels of quenched disorder .
rosch@xcite introduced this model in an effort to understand the temperature dependence of the resistivity exponent measured for the f - electron metal cepd@xmath1si@xmath1.@xcite the results were , in this case , not entirely conclusive because of the anisotropic and complex nature of the spin fluctuation spectrum in this material .
here we present a test of rosch s model in a d - electron system with a cubic structure and @xmath34 one to two orders of magnitude greater than in typical heavy f - electron systems .
we compare the temperature dependence of @xmath16 of two closely related systems on the border of antiferromagnetism .
the two materials are nis@xmath1 near 76 kbar and nis@xmath4se@xmath5 for @xmath6 where the nel temperature @xmath39 vanishes at ambient pressure .
due to random variations in the locations of the s and se atoms in the lattice , values of @xmath28 found in nisse are typically one order of magnitude higher than in stoichiometric nis@xmath1 compounds .
nis@xmath1 crystallizes in the cubic pyrite structure and is an antiferromagnetic insulator at ambient pressure at low temperatures.@xcite it can be metallized via the application of pressure@xcite or by se substitution.@xcitethe temperature - pressure phase diagram of nis@xmath1 and the temperature - composition phase diagram of nis@xmath4se@xmath5 ( fig .
[ ppnissephdiag ] ) are expected to be similar .
of particular interest for this paper is the boundary of metallic antiferromagnetism that appears ( i ) at @xmath40 , i.e. , for nisse , in temperature - composition phase diagram ( fig .
[ ppnissephdiag ] ) and ( ii ) at @xmath3 in the temperature - pressure phase diagram for stoichiometric nis@xmath1 .
we note that in both cases the quantum critical point for metallic antiferromagnetism arises well beyond the metal insulator transition ( see fig .
[ ppnissephdiag ] for the case where composition is the control parameter ) .
we present a high - pressure study of the temperature dependence of the resistivity of nis@xmath1 in the metallic state , which reveals a critical pressure @xmath41 kbar .
the results at @xmath3 are compared with that of nisse at ambient pressure reported in ref . .
in both materials one observes a non - fermi liquid form of the resistivity that appears in first approximation to be consistent with the predictions of the src model ( inset of fig .
[ ppnissephdiag ] for nisse ) .
however , the resistivity of nis@xmath1 near @xmath3 has an undulating component in its variation with temperature that is not seen in nisse .
this difference in behaviour might arise from the effects of cold and hot spots on the fermi surface as suggested by the model due to rosch discussed above.@xcite
single crystals of nis@xmath1 have been grown by chemical vapour transport using iodine as the transport agent.@xcite the residual resistivity ratio @xmath42 above 50 kbar is about 30 for our samples , compared with about 3 for nisse.@xcite the carrier mean free path of nis@xmath1 is thus expected to be about an order of magnitude higher than in nisse .
pressure was applied by means of a non - magnetic bridgman cell using tungsten carbide anvils .
the cell used is a scaled down version of that designed by wittig.@xcite the cullet diameter of the anvils was 3.5 mm .
the gasket was made of pyrophyllite with a central hole to accommodate the nis@xmath1 sample as well as a pb sample and the steatite pressure - transmission medium .
after compression , the sample space was reduced to about 1.5 mm in diameter and 0.1 mm in thickness .
the pressure was determined from the superconducting transition temperature of the pb sample.@xcite the width of the superconducting transition suggested that the pressure variation over the pb sample was about 10% of the average pressure .
the resistivity was measured via the 4-terminal ac technique .
four 50 @xmath43 m pt leads were passed into the high - pressure region via grooves in the insulating pyrophyllite gasket .
the bare wires were rested on top of the samples and pressed onto the sample surface during pressurization to achieve adequate electrical contacts .
the resistivity measurements were carried out with two different and independent systems , a helium circulation cryostat ( ill orange cryostat ; 1.5 k to 300 k ) in grenoble and an adiabatic demagnetization refrigerator ( 0.04 k to 100 k ) in cambridge .
the latter system had two voltage channels , one with a low temperature transformer and the other with a room - temperature transformer .
the excitation currents were 1 ma and 160 @xmath43a in the orange cryostat and in the adiabatic demagnetization fridge , respectively .
the results obtained with these two experimental systems were consistent with each other where comparisons could be made .
temperature dependence of the resistivity in the metallic state of nis@xmath1 at high pressures .
@xmath44 is @xmath45 , where @xmath28 is the residual resistivity extrapolated to @xmath29 k. the curves are shifted vertically for clarity.,height=226 ] low temperature variation of the resistivity in the metallic state of nis@xmath1 at high pressures .
@xmath44 grows in strength up to 76 kbar ( a ) and weakens gradually above 76 kbar ( b).,height=226 ] low temperature variation of the resistivity in the metallic state of nis@xmath1 at high pressures .
@xmath44 grows in strength up to 76 kbar ( a ) and weakens gradually above 76 kbar ( b).,height=226 ] the @xmath46 coefficient of the resistivity , @xmath47 , and residual resistivity , @xmath28 ( inset ) , of nis@xmath1 in the metallic high - pressure state .
the parameters @xmath47 and @xmath28 are obtained by a fit of the resistivity as indicated in the figure .
@xmath47 is peaked at @xmath2 kbar .
this marks the boundary of the metallic antiferromagnetic state of nis@xmath1.,height=226 ] resistivity vs @xmath46 in the high pressure metallic regime of nis@xmath1 .
the three panels are for different temperature intervals .
a quadratic temperature dependence of @xmath44 is seen only at the lowest temperatures ( below a few k ) and the @xmath46 coefficient of @xmath44 is peaked at @xmath48 kbar as shown in fig .
[ ppnisssqcoeff].,height=207 ] resistivity vs @xmath46 in the high pressure metallic regime of nis@xmath1 .
the three panels are for different temperature intervals .
a quadratic temperature dependence of @xmath44 is seen only at the lowest temperatures ( below a few k ) and the @xmath46 coefficient of @xmath44 is peaked at @xmath48 kbar as shown in fig .
[ ppnisssqcoeff].,height=207 ] resistivity vs @xmath46 in the high pressure metallic regime of nis@xmath1 .
the three panels are for different temperature intervals .
a quadratic temperature dependence of @xmath44 is seen only at the lowest temperatures ( below a few k ) and the @xmath46 coefficient of @xmath44 is peaked at @xmath48 kbar as shown in fig .
[ ppnisssqcoeff].,height=207 ] resistivity vs @xmath17 in the high - pressure metallic regime of nis@xmath1 .
the three panels are for different temperature intervals .
an approximately @xmath17 variation of @xmath44 is seen around @xmath48 kbar only over a decade in temperature above a few k. @xmath44 is quadratic in temperature below a few k at all pressures in the metallic regime studied ( see fig .
[ ppnisssqht]c).,height=207 ] resistivity vs @xmath17 in the high - pressure metallic regime of nis@xmath1 .
the three panels are for different temperature intervals .
an approximately @xmath17 variation of @xmath44 is seen around @xmath48 kbar only over a decade in temperature above a few k. @xmath44 is quadratic in temperature below a few k at all pressures in the metallic regime studied ( see fig .
[ ppnisssqht]c).,height=207 ] resistivity vs @xmath17 in the high - pressure metallic regime of nis@xmath1 .
the three panels are for different temperature intervals .
an approximately @xmath17 variation of @xmath44 is seen around @xmath48 kbar only over a decade in temperature above a few k. @xmath44 is quadratic in temperature below a few k at all pressures in the metallic regime studied ( see fig .
[ ppnisssqht]c).,height=207 ] the temperature variations of the resistivity of nis@xmath1 in the high pressure metallic state above 40 kbar are presented in figures [ ppnissov ] and [ ppnissovltlp ] .
( the insulator - to - metal transition ( not shown ) was observed at around 30 kbar , in agreement with the literature.@xcite ) figure [ ppnissov ] shows @xmath44 vs @xmath15 up to 80 k and figure [ ppnissovltlp ] is an expanded view of @xmath44 vs @xmath15 in the range 0.05 k to 2 k. plots of @xmath44 vs @xmath46 and @xmath44 vs @xmath17 over different ranges in temperatures are shown , respectively , in figures [ ppnisssqht ] and [ ppnissthht ] .
below 1 k the resistivity can be described by an equation of the form @xmath49 over the entire pressure range explored , 43 kbar to 96 kbar .
the pressure variations of the fitted values of @xmath47 and @xmath28 are given in figure [ ppnisssqcoeff ] .
the @xmath46 coefficient @xmath47 exhibits a peak at @xmath2 kbar .
contours of the resistivity exponent @xmath26 in the temperature - pressure plane for nis@xmath1 in the metallic regime . a fermi liquid ( fl ) @xmath46 temperature dependence of the resistivity is seen at the lowest temperatures up to @xmath50 defined , somewhat arbitrarily , by the condition @xmath51 .
@xmath50 reaches a minimum at @xmath48 kbar . a non - fermi liquid @xmath52 form of the resistivity with @xmath53
is seen over approximately a decade in temperature above a few k near @xmath3.,height=226 ] the non - fermi liquid form of the resistivity of nis@xmath1 near @xmath48 kbar .
( a ) @xmath54 vs @xmath15 does not saturate and thus has a non - fermi liquid form over a wide temperature range , except at very low temperatures ( fig .
[ ppnisssqht ] ) .
( b ) a plot of @xmath55 vs @xmath56 and the resistivity exponent @xmath57 vs @xmath56 exhibits an undulating form in contrast to the simple @xmath17 temperature dependence seen in nisse ( ref . and fig . [ ppnissephdiag ] ) .
this undulating form of the resistivity has been seen in different nis@xmath1 specimens and with two independent measurement systems.,height=226 ] the non - fermi liquid form of the resistivity of nis@xmath1 near @xmath48 kbar .
( a ) @xmath54 vs @xmath15 does not saturate and thus has a non - fermi liquid form over a wide temperature range , except at very low temperatures ( fig .
[ ppnisssqht ] ) .
( b ) a plot of @xmath55 vs @xmath56 and the resistivity exponent @xmath57 vs @xmath56 exhibits an undulating form in contrast to the simple @xmath17 temperature dependence seen in nisse ( ref . and fig .
[ ppnissephdiag ] ) .
this undulating form of the resistivity has been seen in different nis@xmath1 specimens and with two independent measurement systems.,height=226 ] figures [ ppnisssqht ] and [ ppnissthht ] compare @xmath44 vs @xmath46 and @xmath44 vs @xmath17 , respectively , in three panels each covering different ranges in temperature .
figure [ ppnisssqht]c , in particular , highlights the @xmath46 variation observed at all pressures with a peak of @xmath47 at around @xmath3 as discussed above .
figures [ ppnissthht]a , [ ppnissthht]b and [ ppnissncontour ] , on the other hand , suggest that near this pressure @xmath44 varies roughly as @xmath17 over a decade in temperature above a few k. this is the behaviour predicted by the scr model for a system on the border of metallic antiferromagnetism as discussed in the introduction .
the existence of a @xmath46 regime even at @xmath3 , however , suggests that the transition into the antiferromagnetic state below @xmath3 may not be continuous , i.e. , that the antiferromagnetic quantum critical point is not quite reached due to the onset of a first order transition .
we note that the weak pressure variation of the fermi liquid crossover temperature @xmath50 near @xmath3 ( see fig .
[ ppnissncontour ] and the caption ) would seem to rule out an explanation of the @xmath46 resistivity in terms of an inhomogeneity in pressure .
the identification of @xmath3 with the boundary of metallic antiferromagnetism is also suggested by the correspondence of the temperature - pressure and temperature - composition phase diagram of nis@xmath1 found in earlier work.@xcitethis predicted that the critical pressure for the border of metallic antiferromagnetism should be around 60 kbar , which is of the same order of magnitude as @xmath3 defined above .
the identification of @xmath3 with the antiferromagnetic boundary in the metallic state is tentative and needs confirmation by other measurements and in particular the detection of a signature of @xmath39 in the resistivity for pressures below @xmath3 .
here we focus attention mainly on the behaviour of @xmath16 near @xmath3 and contrast it with that of nisse at ambient pressure ( inset of fig .
[ ppnissephdiag ] ) .
the non - fermi liquid behaviour of @xmath44 over a wide temperature range near @xmath3 ( low - temperature data at 76 kbar with added 77 kbar data up to 300 k ; the high - temperature data is scaled to match the low - temperature data ) is highlighted in figure [ ppnisspcsqdev ] .
panel ( a ) shows the dramatic upturn of @xmath44/@xmath46 , which in the fermi liquid regime is expected to saturate to a constant value .
panel ( b ) is a plot of @xmath55 vs @xmath56 which shows that the average slope corresponds to a resistivity exponent close to @xmath19 , as discussed above and as seen in nisse at ambient pressure .
however , in contrast to nisse , the temperature dependence of the resistivity in nis@xmath1 at @xmath3 exhibits an undulating structure which is evident in figure [ ppnisspcsqdev ] and highlighted in the temperature dependence of the resistivity exponent @xmath26 shown in the inset .
we note that this undulating structure has been observed in several samples of nis@xmath1 and in two independent measurement systems .
the resistivity measurements suggest that antiferromagnetism in the high - pressure metallic state of nis@xmath1 is suppressed at a critical prssure of @xmath2 kbar . at @xmath3 ,
we find that the @xmath46 coefficient of the resistivity @xmath47 has its maximum and the fermi liquid crossover temperature @xmath50 defined in the caption of figure [ ppnissncontour ] has its minimum . above @xmath50 near @xmath3 the resistivity has an approximately @xmath17 temperature dependence over a decade in temperature .
as already stated , this is the behaviour expected in the scr model for a metal in 3d on the border of antiferromagnetism at low temperatures .
also , we note that @xmath3 is of the same order of magnitude as that inferred for the correspondence between the temperature - composition and temperature - pressure phase diagram as discussed in the previous section .
the weakness or absence of a signature of @xmath39 in the resistivity below @xmath3 is not necessarily surprising since the same is the case for nis@xmath4se@xmath5 in the range @xmath58 ( refs . ) where antiferromagnetic order has been confirmed by neutron scattering.@xcite the signature of @xmath39 in @xmath16 can be washed out by inhomogeneities in composition or pressure and may only show up in high - precision measurements of @xmath59 in the pressure range where @xmath39 is not too strongly varying with pressure and where pressure is hydrostatic . the fact that @xmath50 and correspondingly the @xmath46 coefficient of the resistivity @xmath47 remain finite at @xmath3 is not necessarily inconsistent with our assumption that @xmath3 marks the border of antiferromagnetism .
the antiferromagnetic transition vs pressure may be first order as is found in a number of antiferromagnetic metals such as ymn@xmath1 and gdmn@xmath1.@xcite first order quantum phase transitions also seem to be common in ferromagnetic metals such as mnsi and ni@xmath8al.@xcite@xmath50 and @xmath47 remain finite at the antiferromagnetic boundary of ymn@xmath1 , a system that shares some features in common with nis@xmath1 .
( we note , however , that the absence of fermi liquid behaviour does not necessarily mean that the quantum phase transition is continuous.@xcite ) the fermi liquid crossover temperature , @xmath50 , in nis@xmath1 is very much lower than the characteristic spin fluctuation temperature , @xmath34 , which may be expected to be of the order of @xmath60 k in a typical d - metal .
we consider to what extent the behaviour above @xmath50 , but below @xmath34 , may be understood in terms of an itinerant - electron model for a continuous antiferromagnetic quantum critical point .
the predictions of the scr model as analyzed by rosch has already been outlined in the introduction for an antiferromagnetic quantum critical point . with increasing reduced temperature @xmath33
, the scr model predicts that the resistivity exponent @xmath26 drops from an initial value of 3/2 towards unity around @xmath36 , back to a value of order 3/2 around @xmath37 and then to zero for @xmath61.@xcite this behaviour is qualitatively similar to that seen in nis@xmath1 at @xmath3 and above @xmath50 ( inset fig .
[ ppnisspcsqdev ] ) . for reasonable choices of parameters for nis@xmath1 , @xmath62 k , @xmath63 @xmath64 and @xmath65 @xmath64,@xcite we expect the minimum of @xmath30 to occur near 15 k and the maximum at around 120 k. these crossover temperatures are in rough agreement with our observations . in nisse
, @xmath32 is an order of magnitude smaller than in nis@xmath1 and thus the minimum and maximum of @xmath30 are expected to arise at around 100 k and 320 k , respectively . over the temperature range of the experiments shown in the inset of figure [ ppnissephdiag ] the model predicts a simple @xmath17 temperature dependence without modulation , as is seen .
the difference in the behaviour of nis@xmath1 and nisse at their respective critical conditions is thus not surprising .
the combined effects of scattering from spin and lattice fluctuations might also lead to an undulating behaviour of @xmath30 , but this would naively be expected to arise in both nis@xmath1 and nisse , in contradiction with observation .
the temperature dependence of the resistivity of nis@xmath1 near the critical pressure , which has been found to be @xmath2 kbar , is consistent with that expected for a metal on the border of itinerant - electron antiferromagnetism .
the fermi liquid crossover temperature , @xmath50 , which defines the range over which the resistivity is roughly quadratic in temperature , does not vanish at @xmath3 , but is three orders of magnitude smaller than the characteristic spin fluctuation temperature @xmath34 . the finite value of @xmath50 may indicate that the antiferromagnetic quantum critical point is first order , as is the case in related materials such as ymn@xmath1 . over a wide temperature range above @xmath50
the temperature variation of the resistivity exhibits a non - fermi liquid form that can be understood in terms of the effects of spin - fluctuation scattering at cold and hot spots of the fermi surface as anticipated by rosch in his refined treatment of the self - consistent - renormalization ( scr ) model .
in particular , the resistivity exponent @xmath66 exhibits an undulating structure which is consistent not only qualitatively , but also approximately quantitatively with the predictions of this model .
the model also accounts for the absence of this undulating structure in the related , but less pure material , nisse , which is known to be at the border of antiferromagnetism at ambient pressure . | we report low - temperature and high - pressure measurements of the electrical resistivity @xmath0 of the antiferromagnetic compound nis@xmath1 in its high - pressure metallic state .
the form of @xmath0 suggests that metallic antiferromagnetism in nis@xmath1 is quenched at a critical pressure @xmath2 kbar . near @xmath3
the temperature variation of @xmath0 is similar to that observed in nis@xmath4se@xmath5 near the critical composition @xmath6 where the nel temperature vanishes at ambient pressure . in both cases
@xmath0 varies approximately as @xmath7 over a wide range below 100 k. however , on closer analysis the resistivity exponent in nis@xmath1 exhibits an undulating variation with temperature not seen in nisse ( @xmath6 ) .
this difference in behaviour may be due to the effects of spin - fluctuation scattering of charge carriers on cold and hot spots of the fermi surface in the presence of quenched disorder , which is higher in nisse than in stoichiometric nis@xmath1 . |
the evolution of massive stars will depend upon the initial metallicity of the gas out of which they form , and thus we can expect differences in the relative numbers of various stages of massive stars among nearby galaxies .
( for a comprehensive review of the subject , see maeder & conti 1994 . )
the primary effect that metallicity has is due to its influence on radiatively - driven stellar winds and the resulting mass loss .
typical mass - loss rates for galactic o - type stars are 0.520@xmath4 year@xmath1 ( puls et al .
1996 ) , with the more massive stars losing a greater fraction of their mass during their main - sequence lifetimes depends upon the luminosity @xmath5 as @xmath6 ( pauldrach , puls & kudritzki 1986 ; de jager , nieuwenhuijzen , & van der hucht 1988 ; kudritzki & puls 2000 ) , and since luminosity depends upon mass @xmath7 as @xmath8 for high mass stars ( massey 1998a , using the schaller et al .
1992 @xmath9 evolutionary tracks ) , we expect that the mass - loss rates will depend upon the mass roughly as @xmath10 .
the main - sequence lifetime @xmath11 is a relatively weak function of the mass for high - mass stars , and inspection of the schaller et al .
( 1992 ) @xmath9 tracks suggests that @xmath12 .
so we expect that the total mass loss during the main - sequence phase ( @xmath13 ) will go roughly as @xmath14 .
thus the fractional mass lost , @xmath15 , will go as @xmath16 . and
, this is just on the main - sequence ! stars with luminosities above @xmath17 will suffer episodes of enhanced mass loss as their luminosities exceed the eddington limit once line opacities are taken into account ; this stage is likely identified with the luminous blue variable phase ( stars such as @xmath18 car and s dor ) , and accounts for the humphreys & davidson ( 1979 ) upper luminosity limit in the h - r diagram ( lamers 1997 ) . ] . a very high mass star ( 100@xmath19 )
might then lose 50% of its mass during its evolution , which has a profound effect on its path in the h - r diagram , as first shown by de loore , de grve , & lamers ( 1977 ) , de loore , de grve , & vanbeveren ( 1978 ) , chiosi , nasi , & sreenivasan ( 1978 ) , chiosi , nasi , & bertelli ( 1979 ) , brunish & truran ( 1982 ) , and subsequent investigations .
mass - loss rates will scale with metallicity @xmath20 to some power , with the exponent variously estimated from 1.0 to 0.5 ( abbott 1982 ; lamers & cassinelli 1996 ; kudritzki et al .
1989 ; puls , springmann , & lennon 2000 ; kudritzki & puls 2000 ; vink , de koter , & lamers 2001 ; kudritzki 2002 ) . beyond the main - sequence mass - loss rates are highly uncertain ;
for instance , mass loss during the lbv phase is highly episodic and large , with little agreement in what drives the outbursts ( humphreys & davidson 1994 , maeder & conti 1994 ) .
large uncertainties also exist in the mass - loss rates during the red supergiant ( rsg ) phase , making the subsequent tracks even less certain ( salasnich , bressan , & chiosi 1999 ) .
it is commonly assumed that mass - loss rates for wolf - rayet stars ( wrs ) are independent of initial metallicity , since their atmospheres have been so enriched by the products of their own nuclear burning ( e.g. , schaller et al .
1992 ) , but crowther et al .
( 2002 ) have recently argued that iron is an important element in driving the wr winds , and hence that there will be some @xmath20 dependence in their mass - loss rates .
in addition to the effects of mass loss , stellar evolutionary tracks are also sensitive to the treatments of convection and mixing ( maeder & meynet 1987 ) , and there is considerable disagreement among the pundits as to the proper way to include these in the models ( maeder & conti 1994 ) .
recent emphasis has been on the role that rotation plays in mixing in massive stars ( maeder & meynet 2000 , 2002 ) .
convection and mixing also show some dependence on the metallicity ( see , for example , meynet & maeder 2002 ) , and the uncertainties in their treatment underscores the fact that the physics of massive star evolution is not perfectly well understood at present . in order to advance our understanding of massive star evolution ,
it is necessary to have a solid observational database with which the predictions of stellar evolutionary theory may be compared and refined .
a well - known example is the relative number of blue and red supergiants , which van den bergh ( 1973 ) first suggested varied among nearby galaxies due to the effects of metallicity on massive star evolution .
particularly sensitive tests include the relative numbers of different types of evolved massive stars , such as the relative number of different types of wolf - rayet stars ( wc - type and wn - type ) , or the relative number of rsgs and wrs .
maeder , lequeux , & azzopardi ( 1980 ) proposed that the latter number ratio would be particularly sensitive to metallicity effects .
however , there are many observational difficulties in determining such statistics reliably . for wolf - rayet stars , there are selection effects against finding wn - type wrs ( armandroff & massey 1985 , massey & johnson 1998 ) . for red supergiants ,
the problem is that when we look towards a galaxy such as m 31 or the magellanic clouds we see not only the _ bona fide _
extragalactic rsgs but also foreground galactic red dwarfs in the same color and apparent magnitude range .
massey ( 1998b ) found that _
photometry helped separate rsgs from foreground dwarfs , but was not by itself sufficient .
spectroscopy allows an accurate assessment , however .
although the luminosity indicators for late - type stars are rather subtle , an effective technique is to use the near - ir ca ii triplet lines to determine a star s radial velocity . for many local group galaxies
this provides a very clean separation of foreground red dwarfs from extragalactic red supergiants .
here we extend this technique to our nearest galactic neighbors , the magellanic clouds ( mcs ) .
massey ( 2002a ) estimated the degree of foreground contamination would be about 10% in the appropriate magnitude / color range , far lower than the @xmath21 50% found in m31 , m33 , and ngc 6822 by massey ( 1998b ) , both because the clouds are nearer and at higher galactic latitude .
however , an accurate census of the rsg population in the magellanic clouds is of particular interest , as these galaxies are sufficiently close that a great deal is already known about their blue supergiant population , for which much spectroscopy has been carried out ( massey et al .
1995 , massey 2002a ) .
throughout this paper we will adopt the distance and average reddenings listed by van den bergh ( 2000 ) , namely @xmath22 and @xmath23=0.13 for the lmc , and @xmath24 and @xmath25 for the smc .
our sample of red supergiant candidates comes from a recent _ ubvr _ ccd survey covering most of the clouds ( massey 2002a ) .
the sample was chosen to include potential k- and m - type supergiants , based upon the criteria of @xmath26 and a @xmath27 cutoff such that @xmath28 given the observed @xmath29 color and assumed average reddening ( i.e. , tables 9a and 9b of massey 2002a ) .
a few additional red stars which did not quite meet these requirements were also included .
our spectroscopy used the hydra fiber positioner ( barden & ingerson 1998 ) on the blanco 4-m telescope at cerro tololo inter - american observatory during the nights of ( ut ) 4 - 6 oct 2001 . on the first two nights , grating 380 ( 1200 lines mm@xmath1 , blaze 8000 ) was used in first order with an rg-610 blocking filter to obtain data at the ca ii triplet lines ( @xmath30 8498 , 8542 , 8662 ) . a site 2k
@xmath31 4k ccd was used unbinned , behind a 400 mm focal length camera on a bench spectrograph , to obtain a dispersion of 0.27 pixel@xmath1 , with a wavelength coverage extending from 8000 to 9000 .
a 200 @xmath32 m slit plate was inserted at the output of the fiber bundle to yield a resolution of 1.2 ( 4.5 pixels ) . on the third night
( and for a small portion of an engineering night that immediately preceded the run ) , we used grating kpgl1 ( 632 lines mm@xmath1 , blaze 4200 ) in first order with no blocking filter to obtain spectra in the blue in order to determine spectral subtypes .
the ccd was binned in the spectral direction by a factor of 2 , to obtain a dispersion of 1.2 pixel@xmath1 , with a wavelength coverage extending from 3900 to 6100 .
no slit plate was used , and the resolution ( set by the size of the fibers ) was approximately 4 ( 3.5 binned pixels ) .
the hydra fiber positioner consists of 138 fibers ( 300@xmath32 m , or 2.0-arcsec in diameter ) that can be accurately positioned in a 40-arcminute diameter field of view at the rc focus of the blanco 4-m .
an atmospheric dispersion corrector is mounted above the focal plane .
the closest fiber - to - fiber spacing is approximately 25 arcsecs .
this proved a good match to the density of rsg candidates in most of our fields .
our observing procedure was to configure hydra at the zenith , and then obtain a short exposure of a quartz - lamp projector flat that could be used for flat fielding and for removing the relative transmissions of each fiber .
( the projector flats were taken for each new configuration to guard against slight flexure changes in the ccd dewar as the liquid nitrogen cryogen evaporates . )
we would then offset to the field position and align the telescope using 3 to 7 field orientation probes " ( bundles of 6 closely spaced fibers ) that had been placed at the coordinates of bright stars within the field ; these would also be used for guiding .
our program observations then consisted of 3 exposures of 5 mins in length .
these would be followed by a short exposure of a comparison lamp of he , ne , and ar for wavelength calibration .
we would then return to zenith and reconfigure for the next field .
the observations were all carried out by k.a.g.o . , while p.m. kibitzed from his office in flagstaff using the internet to help examine the data in real time . conditions were relatively good throughout the run , with two hours lost at the beginning of the first night due to fog and the last hour of the third night lost to clouds .
the seeing was poor on the first night ( 3 arcsecs ) but was significantly better ( 1 - 2 arcsecs ) on subsequent nights .
the variation in throughput caused by the changes in seeing have no effect on our results , as we are concerned only with the relative strengths and positions of absorption features , and not on absolute spectrophotometry .
all told , we were able to obtain radial velocity information for 6 fields in the smc and 10 fields in the lmc , with a repeat of one of the smc fields on the second night as a consistency check .
the same 6 smc fields were observed for the purposes of spectral classification , along with 7 of the lmc fields . on our two radial velocity nights we obtained a total of 10 observations ( 5 per night ) of 4 radial velocity standards , spread throughout the night .
several different fibers were used for the standards , and the stars were chosen from the list of standard radial velocity ( rv ) stars in the 2001 _ astronomical almanac _ , selected for being of late - type and accessible during our run .
the stars included hd 12029 ( k2 iii ,
rv=+38.6 km s@xmath1 ) , hd 80170 ( k5 iii - iv , rv=0.0 km s@xmath1 ) , hd 213947 ( k2 , rv=+16.7 km s@xmath1 ) , and hd 223311 ( k4 iii , [email protected] km s@xmath1 ) .
as we will describe in section [ sec : rvs ] there was no systematic difference from one night to the next , and our precision was sufficiently high to detect small inconsistencies in the relative velocities of the standards . for the purposes of spectral classification ,
spectral standards were taken from the list of morgan & keenan ( 1973 ) , and included hd 160371 ( k2.5
ib ) , hd 52005 ( k3 ib ) , hd 52877 ( k7 ib ) , hd 42475 ( m0-m1 ib ) , hd 42543 ( m1-m2 ia - ib ) , hd 36389 ( m2 iab - ib ) , hd 190788 ( m3- ib ) , and hd 89845 ( m4.5 ia ) . after basic ccd processing ( overscan bias subtraction and trimming of the data )
, we reduced the spectra using the iraf dohydra " script .
the quartz - lamp projector flats were used to define the identification , location , and shape of the fiber profiles on the chip .
this information was used to optimally extract " the program objects and comparison exposures ; flat - fielding and removing the fiber - to - fiber variations was done using the extracted projector flat exposures as well .
we found that the illumination of the outlying fibers with the projector flat did not match the sky illumination very well . in sky - limited applications
this would compromise the sky subtraction unless corrected for by observations of blank sky , say , but since our stars were quite bright compared to the sky , this made little difference in our final data .
the extracted spectra were then wavelength calibrated using the extracted comparison - line spectra .
finally , the three one - dimension spectra of each object were averaged using bad - pixel rejection .
the standard star data were treated identically , except that a single exposure was involved and so no averaging was done .
the spectra in the red ( that would be used for radial velocity measurements ) were then normalized by a low - order cubic spline , and then shifted by unity to make the average continuum level zero .
radial velocities were measured by cross - correlating each magellanic cloud spectrum against each of the radial velocity standard star observations . since we could find no systematic effect in cross - correlating the spectra of the radial velocity standards from one night to the next , we simply treated all of our data the same , regardless of which night they were obtained on .
we used the iraf routine `` fxcor '' , and limited the cross correlation to the region 8450-8700 in order to isolate the ca ii triplet ( @xmath30 8498 , 8542 , 8662 ) .
the cross - correlation peaks were fit by a parabola , resulting in an internal precision of 0.5 - 0.7 km s@xmath1 for each measurement . the measurement based on each of the ten standard star observations
were then averaged ; the agreement between these were excellent , and the resulting means had a standard deviation of the mean of 0.2 - 0.3 km s@xmath1 .
we believe this is an honest estimate of our actual accuracy , as quite a few stars were observed twice ( or even three times ) owing either to their locations in overlapping fields , or due to two observation of the same fields on different nights .
the typical agreement for these stars was 0.25 km s@xmath1 .
our spectra are so well exposed , and the ca ii triplet lines so strong , that we could easily detect small systematic differences in the cross - correlations produced by different standard stars . for instance , each of the two observations of the standard star hd 213947 ( obtained on separate nights ) produced cross - correlations that were @xmath34 km s@xmath1 high compared to that obtained from the ensemble , while the standard star hd 12029 produced cross - correlations that were @xmath34 km s@xmath1 low compared to that obtained from the ensemble ) and hd 223311 ( @xmath35 km s@xmath1 ) as correct , then the true radial velocity of hd 213947 is 15.0 km s@xmath1 rather than the 16.7 km s@xmath1 adopted by the iau , while that of hd 12029 is 39.6 km s@xmath1 rather than the 38.6 km s@xmath1 adopted by the iau . ] .
altogether , we obtained radial velocities for 118 stars in the smc .
three were measured three times , and 42 were measured twice . for the lmc , we obtained radial velocities for 167 stars .
of these , seven were measured three times , and 35 were measured twice . in tables
[ tab : smc ] and [ tab : lmc ] we give the average radial velocities of the stars in our sample .
figure [ fig : rvhists ] shows the histograms of these velocities , with the center - of - mass systemic velocities of the clouds indicated . for both the smc and lmc
there is excellent agreement in the peaks of the histograms and the cataloged systemic velocities of each cloud .
the tail " of velocities extending to lower radial velocities is readily identified as the foreground red dwarfs which we had hoped to distinguish from the members of the clouds . for the lmc diagram
the separation is quite clean .
the lower systemic velocities of the smc results in there being a little uncertainty for three stars with intermediate velocities .
we have assigned membership in the tables based upon whether or not the radial velocity is greater than 100 km s@xmath1 . based upon the radial velocities , we conclude that 11.0% of the stars in the smc sample proved to be foreground stars , while only 5.3% of the stars in the lmc sample were foreground stars . we include our spectral types in tables [ tab : smc ] and [ tab : lmc ] .
not all stars were observed in the blue , and hence there are stars for which there are no spectral types .
these were determined by comparison of our spectra of spectral standards to the program objects .
at our dispersion and signal - to - noise , the presence of the tio @xmath36 suggests that the star is k5 or later , and the classification was based upon the strength of the tio bands at @xmath30 4761 , 4954 , 5167 , 5448 , and 5847 .
if there was no tio present , then the relative strength of ca i @xmath37 and the g band were used to determined the spectral subtype in the range k0-k5 .
strong h@xmath38 suggested an earlier type ( g - type ) , which proved to be the case for a few of the foreground dwarfs .
the luminosity criteria are quite subtle , and we relied upon our radial velocities to guide us in assigning v " for foreground dwarfs , or
i " for supergiants .
the presence of lmc giants in our sample is precluded by our @xmath27 magnitude selection criterion . the comparison with the published spectral types for some stars in common with elias , frogel , & humphreys ( 1985 ) [ smc ] , and humphreys ( 1979 ) [ lmc ] shows generally excellent agreement .
the average difference is less than half a spectral type for the 76 stars in common . in only three cases
the difference is 3 spectral subclasses or more ; i.e. , smc-026778 , which we call k2 i but elias et al .
( 1985 ) call m0 i ; smc-054708 , which we call k0 i but elias et al . ( 1985 ) call m0 iab ; and lmc-178066 , which we call k7 i but humphreys ( 1979 ) call m2 ia .
given the size of the discrepancy , we speculate that these may be spectrum variables .
( in the case of the lmc star , the identification is not certain , as only approximate coordinates had ever been published for the case stars that were subsequently observed by humphreys 1979 . )
what do these spectral types mean in terms of physical parameters ?
we have classified the stars in the traditional way , relying upon the strengths of the tio bands to determine the spectral type , with stronger bands leading to a later type . however , the metallicity ( as judged from the oxygen abundances of h ii regions ) of the lmc is about a factor of 2 lower than in the solar neighborhood , while the metallicity of the smc is about a factor of 4 times lower ( russell & dopita 1990 ) .
thus rsgs of the _ same _ effective temperatures in the milky way , lmc , and smc would be classified as progressively earlier in these three galaxies as lower metal abundance weakens the tio band strength used to classify these stars .
elias et al . ( 1985 ) see such an effect in their comparison of the average ( median ) spectral types of rsgs in smc ( m0
i ) , the lmc ( m1 i ) , and the milky way ( m2 - 3 i ) , but attribute the change primarily to the effect that metallicity has on the location of the ( giant branch ) hayashi track , only secondarily to the effects on the spectral appearance of stars of a given effective temperature
. however , modern evolutionary models do not show a hayashi track for red _ super_giants , and so it is worth re - examining this issue .
we show our own histograms for the lmc and smc supergiants in our sample , and compare these to the distribution of spectral types for the milky way taken from table 20 in elias et al .
( 1985 ) . the medians we find are k5 - 7 i for the smc , m1 i for the lmc , and m2 i for the milky way .
the median spectral type in the smc is somewhat earlier than that found by elias et al .
( 1985 ) , and probably results either from our larger sample size , or our better completeness for early k - type supergiants .
we find , as do elias et al .
( 1985 ) , that the distribution of spectral types is more narrow in the smc than in the lmc or milky way , although we still find rsgs as late as m3 i in the smc just not in large numbers .
first , let us ask if it is reasonable that this progression in average spectral types is due solely to the effect that metallicity has on the relationship between spectral type and effective temperature . in table [ tab : teff ] we compare various effective temperature scales for galactic rsgs .
we include here the effective temperature scale adopted by humphreys & mcelroy ( 1984 ) , based upon a number of sources , and the lee ( 1970 ) calibration of effective temperature with spectral types for m - type supergiants , based primarily on a very limited amount of fundamental " data ( i.e. , using stars with known radii ) .
this work has been extended considerably in recent years by dyck et al .
( 1996 ) and dyck , van belle , & thompson ( 1998 ) , who obtained new interferometric observations at 2.2@xmath32 m , and combined these with similar data from the literature .
they provide a scale for red _ giants _ , but consider the supergiant data to be too sparse for a calibration .
their supergiant data clearly lies several hundred degrees cooler than the giant sequence ( i.e. , fig . 3 in dyck et al .
houdashelt et al .
( 2000 ) recently compared the temperatures expected from the new marcs models to the dyck et al .
( 1996 ) values and found very good agreement .
we have adjusted the dyck et al .
( 1996 ) scale for red _ giants _ by @xmath39k ( i.e. , to cooler temperatures ) to produce reasonable estimates for _ supergiants _ , consistent with the temperature differences illustrated in their fig . 3 .
( see also discussion following bessell 1998 . ) comparing all of these values have led to a somewhat arbitrary effective temperature scale which we adopt here , noting that the present uncertainties prevent a more definitive answer at this time .
what sort of change is expected on the basis of metallicity ? improved stellar atmospheres applicable to rsgs are under construction ( gustafsson et al .
2003 , plez 2003 ) but until these are generally available we turn to the atlas 9 " model atmospheres of kurucz ( 1992 ) to help answer this question .
although oestreicher & schmidt - kaler ( 1998 ) find that the atlas 9 models significantly underestimate the amount of molecular absorption for _ some _ lines in late - type stars , we show that there is pretty good agreement in fig .
[ fig : kurucz ] , where we compare the coolest of kurucz ( 1992 ) models to the spectra of three of our spectral standards .
the kurucz ( 1992 ) models correspond to solar metallicity and @xmath40 , which is appropriate for a massive supergiant
. will vary from about @xmath41 ( 20@xmath19 , @xmath42 , @xmath43 ) to @xmath44 ( 40@xmath19 , @xmath45 , @xmath46 ) .
thus the @xmath40 kurucz ( 1992 ) models are the most appropriate ones available for rsgs .
fortunately , the strengths of the tio bands in general are not sensitive to the exact choice of @xmath47 ; see schiavon & barbuy ( 1999 ) .
] we see that the 3500@xmath2k model shows tio lines that are roughly comparable with what is seen in m0-m2 i supergiants , consistent with the effective temperature scale we adopted in table [ tab : teff ] .
similarly , the tio bands in the 4000@xmath2k model are similar to that of the k2.5 i standard , also in accord with the effective temperature scale adopted above .
the spectra are plotted in log units in order to facilitate comparison of band depths without the subjective task of normalization .
the continuum fluxes of the stellar spectra have been adjusted by comparison with stars of similar spectral types from the jacoby , hunter , & christian ( 1984 ) atlas , and so the relative fluxes are only approximate ; what matters is the line depths .
we next investigate the effect the kurucz ( 1992 ) models predict for a change in metallicity from galactic to that of the smc , where we have observed the spectral types change from m2 i to k5 i. the red curve in fig .
[ fig : kurucz ] shows the kurucz ( 1992 ) model for a 3500@xmath2k supergiant computed with an abundance @xmath48 , while the blue curve corresponds to @xmath49 .
the metallicity of smc should be intermediate between these two values .
we see that metallicity alone has changed the line depths to be intermediate between the 3750@xmath2k and 4000@xmath2 models .
thus the change in metallicity from the milky way to that of the smc is likely to weaken the appearance of the tio spectral lines by an amount corresponding to @xmath50k to @xmath51k .
this is consistent with the @xmath52k temperature difference between ( galactic ) m2 i and k5 - 7 i stars .
thus the effective temperature scale at lower metallicity will be cooler ; i.e. , an smc m0 i star would be 300 cooler than a galactic m0 i star .
a more quantitative comparison requires higher resolution synthetic spectra and a finer temperature and metallicity grid , and these will soon be available for such tests from the marcs group ( gustafsson et al .
2003 , and plez 2003 ) that expected for a supergiant , so the application to the stars we discuss here is not straight - forward .
we will note that his spectra apply a considerably warmer temperature scale ( and considerably more compressed ) than what we adopt here . at first blush
, the warmer scale appears to be in disagreement with the fundamental data of lee ( 1970 ) and dyck et al .
the models do not show much of an effect with metallicity , but a more careful comparison done with absolute spectrophotometry , with more appropriate surface gravities , is needed . ] . in the meanwhile , we will adopt an effective temperature scale for the magellanic cloud rsgs that is 300@xmath2 cooler for the smc , and 100@xmath2 cooler for the lmc , compared with the milky way , consistent with the average change in spectral type we observe .
we can provide an additional check on this by examining the intrinsic colors .
it is generally recognized that @xmath53 is a good effective temperature indicator for cool stars , while @xmath54 is sensitive both to effective temperature and to surface gravity ( e.g. , lee 1970 , massey 1998b , oestreicher & schmidt - kaler 1999 ) . in table
[ tab : kurucz ] we give the expected @xmath54 and @xmath53 colors as a function of effective temperature and metallicity computed from the kurucz ( 1992 ) atlas 9 models , where we have adopted the description of the _ b _ and _ v _ band - passes from buser & kurucz ( 1992 ) , and that of the kron - cousins _ r _ bandband from bessel ( 1983 ) .
we include in table [ tab : kurucz ] the approximate corresponding spectral types for galactic stars , using table [ tab : teff ] .
we see that there is very little change in color with metallicity for the warmer " models ( 3750@xmath2k to 4000@xmath2k ) . for the 3500@xmath2k model
there is no change from the milky way to the lmc , but that we expect that @xmath53 will be significant larger ( 0.07 mag ) in the smc .
how do these colors compare to the observed photometry ? in table [ tab : colors ] we give the average @xmath54 and @xmath53 colors for our spectral types , where we have corrected the observed colors in tables [ tab : smc ] and [ tab : lmc ] by the average reddenings as indicated in the footnote to the table .
we have used the arithmetic means at each spectral type , after rejecting the highest and lowest values in producing these averages .
we do not list colors for any spectral types with 3 or fewer representatives . for the lmc
there is relatively good agreement : we expect an lmc m0 i star to have @xmath55k ( i.e. , 100@xmath2 cooler than the value listed in table [ tab : teff ] ) and the kurucz ( 1992 ) model atmospheres predict a @xmath53 color of 0.92 .
we observe @xmath56 ( table [ tab : colors ] ) .
however , for the smc the agreement is poor between the kurucz ( 1992 ) @xmath53 colors and those observed .
if we correct the galactic scale by @xmath57k as argued above , a 3500@xmath2k smc star should have a spectral type of k5 i. the models then predict a @xmath53 color of 0.99 .
but , what we actually observe is a @xmath53 color of 0.84 .
if we had made _ no _ correction to the galactic effective temperature scale then the broad - band colors would be in pretty good agreement .
have we fooled ourselves in making this correction ? possibly
. however , bessell et al . (
1989 ) has published a few models applicable to cool supergiants .
we give their @xmath53 colors in table [ tab : bessell ] .
their smc - like metallicity ( @xmath58 ) supergiant model predict a @xmath53 color of 0.84 at 3500@xmath2k ( @xmath59g=-0.26 ) , in excellent agreement with the observed colors if we apply our temperature correction . similarly , their @xmath60k model predicts a @xmath53 color of 0.92 . applying our correction
, we would expect this temperature would correspond to an smc star of spectral type k7-m0 i , and indeed we find an observed color of 0.88 - 0.94 , in good agreement .
lmc - like metallicity models were not computed by bessell et al .
( 1989 ) , limiting the degree we can make this comparison , but we note that their galactic @xmath53 colors are significantly bluer than the kurucz ( 1992 ) models would predict ( e.g. , 0.74 vs. 0.92 at @xmath60k ) . a finer grid of higher - resolution synthetic spectra appropriate to cool supergiants is needed before a metallicity - dependent effective temperature scale can be reliably derived .
let us next compare the distribution of stars in the h - r diagram to that predicted by stellar evolutionary models .
we use the corrected " temperatures for the spectral types , as defined above . for stars without spectral types , we can use the @xmath53 to determine an effective temperature .
comparison of our measured colors for the stars with spectral types produces two linear relations : @xmath61 @xmath62 the conversion from the adopted effective temperatures to bolometric corrections is made by using the relation of slesnick , hillenbrand , & massey ( 2002 ) , which is primarily a fit to the bolometric corrections as a function of effective temperatures tabulated by humphreys & mcelroy ( 1984 ) .
we have included the adopted @xmath63 and @xmath64 in tables [ tab : smc ] and [ tab : lmc ] . as the massey ( 2002a )
photometric survey was limited in area and could conceivably suffer from saturation for the most luminous supergiants , we should also consider other stars that have been spectroscopically confirmed as magellanic cloud rsgs .
we list these in tables [ tab : smcother ] and [ tab : lmcother ] . in the case of the smc we have excellent cross - reference to the spectral types of elias et al .
( 1985 ) thanks to the good coordinates provided by sanduleak ( 1989 ) . however , cross - referencing to the spectral types of humphreys ( 1979 ) was more of a challenge as only approximate coordinates were provided in the case objective prism survey ( sanduleak & philip 1977 ) from which humphreys ( 1979 ) drew her sample for spectroscopy . thus massey ( 2002a ) gives all cross - identifications for the lmc stars as tentative , and there were a number of stars for which several possible matches were a possibility .
thus there may be other previously observed lmc rsgs that we have incorrectly adopted as identical to our stars in table [ tab : lmc ] .
we show the h - r diagrams in figs .
[ fig : smchrds ] and [ fig : lmchrds ] , where we include the evolutionary tracks both of the geneva and padova groups ( schaerer et al .
1993 ; meynet et al . 1994 ; fagotto et al .
we see here a very interesting effect , namely that _ none _ of these evolutionary tracks produce rsgs that are as cool and as luminous as observed in the magellanic clouds .
although the agreement is good at 12@xmath19 masses , at higher masses the tracks simply do not go far enough to the right ( cool temperatures ) to produce the rsgs that we observe .
it would appear that the rsg sequence extends up to perhaps 40@xmath19 , but that those tracks simply do not go sufficiently far to the right in the diagram .
massey ( 2002b , 2003 ) finds that the identical problem exists for galactic rsgs even when adopting the effective temperature scale and luminosities of humphreys ( 1978 ) .
this discrepancy has also been suggested by the poor match of synthetic starburst " spectra to observations of the integrated light of various stellar populations ( e.g. , mayya 1997 , oliva & origlia 1998 , origlia et al .
it is quite apparent even if one looks only at the broad - band colors derived by lejeune & schaerer ( 2001 ) .
for instance , if one considers the 40@xmath19 evolutionary track computed with @xmath65 and a normal mass - loss rate by charbonnel et al .
( 1993 ) , which is shown in fig .
4(a ) , lejeune & schaerer 2001 ) compute @xmath66 and @xmath67 at the coolest extension of the track .
these colors correspond to a mid - f type supergiant . what can account for the problem with the evolutionary tracks ?
one possibility is the difference that the treatment of convection can make in the evolutionary tracks .
this is illustrated in fig .
9 of maeder & meynet ( 1987 ) , where they compare the older bhm - vitense ( 1981 ) mixing length ( 1.5 times the local pressure scale height ) with a more accurate treatment that includes the effects of turbulent pressure and acoustic flux and has the mixing length proportional to the density scale height .
although the physics is better , the result is that the evolutionary tracks no longer produce rsgs that are as luminous and cool as earlier models had .
however , the padova models reply upon the older bhm - vitense ( 1981 ) prescription , albeit it with a mixing length of 1.63 times the pressure scale height , and these too suffer from the same problem , as shown in figs . [
fig : smchrds ] and [ fig : lmchrds ] .
maeder & meynet ( 1987 ) were certainly aware of the mis - match between theory and observation , and expressed the hope that
complete stellar models " ( i.e. , ones which included the extended atmospheres caused by stellar winds ) would some day alleviate the problem .
such winds would make the star larger than the ( purely ) interior models would suggest , lowering the effective temperature . in the meanwhile , this discrepancy has been generally ignored by the users of these models . what if we had ignored the effective temperature corrections ? in fig .
[ fig : oldhrds ] we compare the h - r diagrams for the smc and lmc adopting the galactic spectral type to effective temperature calibration from table [ tab : teff ] , and computing the corresponding color to effective temperature equation .
it is clear that if we made no correction that the rsgs would be of considerably higher luminosity in the lmc than in the smc . in the left - hand side of fig .
[ fig : mbolhists ] we compare the distribution of bolometric luminosities for rsgs both with and without these corrections .
we see that without the corrections the number of rsgs in the smc drops very abruptly with increasing luminosity compared to the lmc .
this runs counter to the expected evolutionary effect that at lower metallicities higher mass ( luminosity ) stars should spend a greater fraction of their he - burning lifetimes as rsgs rather than wrs since mass - loss rates will be lower at low metallicities ( see maeder et al .
1980 and maeder & conti 1994 ) . indeed ,
massey ( 1998b ) found a smooth decrease in the numbers of the highest luminosity rsgs as metallicity increased from ngc 6822 ( smc - like ) to m 33 ( lmc - like ) to m 31 ( higher than solar ) .
when we make the correction for effective temperature , however , the luminosity functions become very similar ( right - hand side of fig .
[ fig : mbolhists ] .
thus either a significant correction to the galactic @xmath63 scale is needed for smc rsgs , as we have made above , or else there is an unexpected absence of higher luminosity rsgs in the smc .
we note that in the case of the h - r diagrams with the corrected temperatures that we expect that the most luminous rsgs come from stars with initial masses of about 40@xmath19 .
this is consistent with so - called upper luminosity limit described by humphreys & davidson ( 1979 ) , and explained by lamers ( 1997 ) : the modified " eddington limit should prevent stars with luminosities above @xmath68 from evolving to the right in the h - r diagram .
this limit should , if anything , be slightly higher at lower metallicities , since the opacities will be lower , and thus is consistent with our corrected temperatures .
perhaps one of the most interesting things to be apparent in the h - r diagrams is that there is a very smooth decrease in effective temperature with increasing luminosity , whether or not the temperature corrections are applied . _
the higher luminosity rsgs are invariably of cooler effective temperatures .
_ this tight sequence is obviously not reproduced by the stellar evolutionary models . explaining this simple sequence provides an important challenge to stellar evolutionary theory .
finally , let us briefly reconsider the ratio of blue - to - red supergiants ( b / r ) in the smc and the lmc .
as emphasized in massey ( 2002 ) , one needs to be careful in what one is counting for this ratio to have much meaning .
we need to include k - type as well as m - type stars , but would like to exclude stars earlier than k - type as the degree of foreground contamination increases at intermediate colors .
we adopt the same convention as massey ( 2002a ) , namely @xmath26 , corresponding to a star of @xmath69 ( 4600@xmath2k , or late g - type ) .
we also restrict ourselves to counting only stars with @xmath70 , as less luminous than this there is a chance of contamination by intermediate - mass asymptotic giant branch stars ( brunish , gallagher , & truran 1986 ) . in counting stars
we include all of the sufficiently luminous rsgs in tables [ tab : smc ] and [ tab : lmc ] , plus a fraction of the other red stars from massey ( 2002a ) .
our spectroscopy suggests that 11% of the red stars seen towards the smc , and 5.3% of the red stars seen towards the lmc are foreground , so we count only 89% and 94.7% of the remainder and @xmath71 , we started with the complete photometric catalog ( table 3 ) of massey ( 2002a ) rather than the list of just the red , luminous stars ( table 9 ) , but the differences are small . ] .
we find that we expect about 90 rsgs in the smc sample , and 234 rsgs in the lmc sample .
for the blue stars , we use the numbers given in table 10 of massey ( 2002 ) , i.e. , all of the stars in the smc and lmc areas surveyed that meet the criteria @xmath70 and @xmath72 , where the latter roughly corresponds to the color of an a9 i star ( @xmath73 ) .
we then count 1484 blue supergiants in the smc , and 3164 blue supergiants in the lmc , although these numbers are considerably uncertain given the difficulty in converting photometry to @xmath71 and @xmath64 for hot stars .
( see , for example , massey et al .
the derived b / r values are thus 16 for the smc , and 14 for the lmc , essentially identical . massey ( 2002 )
notes that a large ( factor of 3 ) difference is found if only m - type stars are counted .
thus the fact that the stellar models fail to reproduce the observed " b / r value ( langer & maeder 1995 ) may be in large part due to the differences in how the observed " ratios have actually been counted .
the slightly different approach here has changed the b / r ratio given by massey ( 2002a ) by nearby a factor of 2 in itself , and thus we again emphasize the large observational " uncertainty in such a census , as the derived ratio is highly sensitive to the conversions to bolometric luminosity .
we can compare this number to the predicted from stellar models .
for this comparison we follow the advice offered by schaerer & vacca ( 1998 ) to determine the number of stars from the model by integrating the initial mass function over closely - spaced isochrones rather than by integrating over the coarsely - spaced mass tracks . the smc - like @xmath65 geneva models with normal mass - loss rates ( charbonnel et al . 1993 )
predict a blue - to - red supergiant ratio of 54 , while the enhanced mass - loss models ( meynet et al .
1994 ) would expect a blue - to - red supergiant ratio of 36 .
the lmc - like @xmath74 geneva models with normal mass - loss rates ( schaerer et al .
1993 ) predict a b / r value of 10 , while those using enhanced mass - loss rates ( meynet et al .
1994 ) predict a b / r value of 3 .
thus , the lower metallicity models ( smc - like ) predict a much higher ratio of b / r than what is observed , while the intermediate metallicity models ( lmc - like ) predict a somewhat lower value . given that we have earlier shown that the models fail to reproduce the location of rsgs in the h - r diagram , the disagreement is not surprising .
maeder & meynet ( 2001 ) find that more rsgs are produced in the models at smc - like metallicities when rotation is included .
comparisons with the new rotation models that cover a range of metallicities will be of great interest .
we have examined samples of red stars seen towards the smc and lmc .
our spectroscopy has been able to determine membership based upon radial velocity information ; we find that the contamination by foreground red dwarfs is about 11% in the smc sample and 5.3% in the lmc sample .
classification of our spectra confirms that there is a progression in the average spectral type of rsgs with metallicity .
rsgs in the smc ( which is the lowest in metallicity ) have an average spectral type of k5 - 7 i. nevertheless , there are a few smc rsgs as late as m3 in our sample . in the lmc
have an average type of m1 i , while those in the milky way have an average type of m2 i. at lower metallicity the appearance of the tio lines ( used as the primary classification criterion ) should be weaker , and examination of the kurucz ( 1992 ) atlas 9 models suggest that this effect is probably sufficient in itself to account for the shift in average spectral types observed .
if so , then rsgs in the smc are about 300@xmath2 cooler than galactic stars of the same spectral types , while rsgs in the lmc are about 100@xmath2 cooler .
the @xmath53 colors predicted by the kurucz ( 1992 ) models do not agree with this conclusion , but other models ( e.g. , bessell 1989 ) show better agreement with the smc data although lack the lmc - like metallicity we would need to draw conclusions . good resolution ( @xmath75 ) synthetic spectra for red supergiants ( @xmath40 ) covering a range of metallicities is needed ( along with good spectrophotometry ) to address this discrepancy .
we find that none of the stellar evolutionary models produce rsgs that are as red and luminous as observed in the magellanic clouds .
this discrepancy may be due to the treatment of convection in the evolutionary models , or could simply be due the lack of inclusion of that stellar winds have in increasing the atmosphere extent ( leading to a decrease in the effective temperature ) in the stellar models .
nevertheless the location of rsgs compared to the evolutionary tracks suggest that the most luminous rsgs have evolved from stars with initial masses of 40@xmath19 , in accord with previous studies .
we show that ignoring the temperature correction described above would lead to an under - abundance of high luminosity smc rsgs .
the blue - to - red supergiant ratio does not appear to be significantly different in the smc and in the lmc , although there is still considerable uncertainty in the number of blue supergiants in our sample .
this work has underscored the point made by massey ( 2002a ) that the b / r value is very dependent upon how stars are counted , and thus disagreements with the predictions of stellar evolutionary models have to be carefully evaluated .
using the non - rotation geneva models , we find that the smc - like ( @xmath65 ) models predict too large a value for b / r , while the lmc - like ( @xmath74 ) models predict too small a value .
given the fact that the models fail to produce high luminosity red supergiants , such disagreements are not surprising .
the effects that rotation will have on the predicted b / r ratio as a function of metallicity remain unclear . as maeder & meynet ( 2000 ) describe , the additional mixing caused by rotational instabilities would tend to produce few rsgs , while on the other hand higher rotation will lead to increase mass loss , and this would tend to produce more rsgs .
the result is that it is still unclear what affect , if any , the complete inclusion of rotation will have on the predictions of b / r ratios as a function of metallicities , although maeder & meynet ( 2001 ) find that at a smc - like metallicity including rotation will lower the predicted b / r ratio , which goes in the correction direction .
eggenberger , meynet & maeder ( 2002 ) compare the observed b / r ratios of clusters to those of models , but as discussed extensively by massey ( 2002 , 2003 ) the b / r " ratio in a quasi - coeval situation will be quite different than in a mixed - age population , such as what we consider here .
this work was supported by nsf grant ast0093060 .
we are grateful , as always , to the kind hospitality and excellent support by the technical staff at ctio .
we also benefited from correspondence with ken hinkle and daniel schaerer .
deidre hunter offered useful comments on an early draft of this manuscript .
gustafsson , b. , edvardsson , b. , eriksson , k. , grae - jorgensen , u. , mizuno - wiedner , m. , & plez , b. 2003 , in stellar atmosphere modeling , ed .
i. hubeny , d. mihalas , & k. werner ( san francisco , asp ) , in press lamers , h. j. g. l. m. , & cassinelli , j. p. 1996 , in from stars to galaxies : the impact of stellar physics on galaxy evolution , ed . c. leitherer , u. fritze - von alvensleben , & j. huchra ( san francisco : asp ) , 162 l c c c c c c r r l l l 008324&00 47 16.84&@xmath3373 08 08.4&13.08 & 1.64 & 0.85 & 3.565 & -7.32 & 134.4&smc & k0 : i & + 008367&00 47 18.11&@xmath3373 10 39.3&12.46 & 1.40 & 0.93 & 3.531 & -8.58 & 127.9&smc & k7
i & + 008930&00 47 36.94&@xmath3373 04 44.3&12.68 & 2.00 & 1.06 & 3.531 & -8.36 & 131.6&smc & k7 i & m1 ia + 009766&00 48 01.22&@xmath3373 23 37.5&12.95 & 1.29 & 0.86 & 3.531 & -8.09 & 141.6&smc & k7 i & + 010889&00 48 27.02&@xmath3373 12 12.3&12.20 & 2.00 & 1.06 & 3.531 & -8.84 & 138.4&smc & k7 i & m0 ia + 011101&00 48 31.92&@xmath3373 07 44.4&13.54 & 1.69 & 0.99 & 3.531 & -7.50 & 146.4&smc & k7 i & + 011709&00 48 46.32&@xmath3373 28 20.7&12.43 & 1.79 & 0.94 & 3.531 & -8.61 & 140.4&smc & k7 i & k5-m0i + 011939&00 48 51.83&@xmath3373 22 39.3&12.82 & 1.81 & 1.00 & 3.518 & -8.59 & 131.8&smc & m0 i & + 012322&00 49 00.32&@xmath3372 59 35.7&12.44 & 1.93 & 1.03 & 3.531 & -8.60 & 149.0&smc & k7 i & m0 ia + 012572&00 49 05.25&@xmath3373 31 07.8&11.66 & 1.45 & 0.76 & 3.602 & -8.35 & 228.5&smc & & + 012707&00 49 08.23&@xmath3373 14 15.5&13.40 & 1.77 & 1.00 & 3.503 & -8.52 & 162.6&smc & & + 013472&00 49 24.53&@xmath3373 18 13.5&11.73 & 1.77 & 0.85 & 3.531 & -9.31 & 137.6&smc & k7 : i & k0-k5i + 013740&00 49 30.34&@xmath3373 26 49.9&13.47 & 1.77 & 0.96 & 3.531 & -7.57 & 156.4&smc & k7 i & + 013951&00 49 34.42&@xmath3373 14 09.9&13.00 & 1.79 & 0.93 & 3.531 & -8.04 & 125.0&smc & k7 i & + 015510&00 50 06.42&@xmath3373 28 11.1&12.59 & 1.90 & 0.95 & 3.518 & -8.82 & 163.0&smc & m0 i & m0
i + 017656&00 50 47.22&@xmath3372 42 57.2&12.66 & 1.69 & 0.90 & 3.568 & -7.70 & 134.0&smc & k0 - 5 i & + 018592&00 51 03.90&@xmath3372 43 17.4&11.39 & 1.82 & 0.95 & 3.568 & -8.97 & 152.3&smc & k0 - 2 i & k5-m0i + 019551&00 51 20.23&@xmath3372 49 22.1&12.98 & 1.04 & 0.83 & 3.568 & -7.38 & 145.4&smc & k2 i & + 019743&00 51 23.28&@xmath3372 38 43.8&13.45 & 1.67 & 1.05 & 3.544 & -7.29 & 138.2&smc & k5 i & m0 iab + 020133&00 51 29.68&@xmath3373 10 44.3&12.33 & 1.95 & 1.03 & 3.518 & -9.08 & 170.4&smc & m0 i & m0 iab + 020612&00 51 37.57&@xmath3372 25 59.5&12.97 & 1.64 & 0.82 & 3.544 & -7.77 & 154.9&smc & k5 i & k5-m0 + 023463&00 52 26.51&@xmath3372 45 15.6&12.44 & 1.35 & 0.90 & 3.568 & -7.92 & 157.9&smc & k0 - 5 i & + 023700&00 52 30.69&@xmath3372 26 46.8&13.09 & 1.67 & 0.85 & 3.568 & -7.27 & 149.8&smc & k0 - 2 i &
+ 025550&00 53 02.85&@xmath3373 07 45.9&13.35 & 1.67 & 0.94 & 3.568 & -7.01 & 136.8&smc & k2 i & + 025879&00 53 08.87&@xmath3372 29 38.6&11.91 & 1.77 & 0.88 & 3.531 & -9.13 & 134.5&smc & k7 i & m0 ia + 025888&00 53 09.04&@xmath3373 04 03.6&12.08 & 1.82 & 0.95 & 3.538 & -8.80 & 159.1&smc & k5 - 7 i & m0 ia- + 026402&00 53 17.81&@xmath3372 46 06.9&12.78 & 1.05 & 0.75 & 3.568 & -7.58 & 148.4&smc & k0 - 2 i & + 026778&00 53 24.56&@xmath3373 18 31.6&12.78 & 1.55 & 0.95 & 3.568 & -7.58 & 153.0&smc & k2 i & m0 iab + 027443&00 53 36.44&@xmath3373 01 34.8&12.75 & 1.86 & 1.01 & 3.531 & -8.29 & 140.3&smc & k7 i & + 027945&00 53 45.74&@xmath3372 53 38.5&12.94 & 1.57 & 0.80 & 3.552 & -7.61 & 135.4&smc & k3 - 5 i & + 030135&00 54 26.90&@xmath3372 52 59.4&12.84 & 1.68 & 0.78 & 3.568 & -7.52 & 150.8&smc & k0 - 2 i & + 030616&00 54 35.90&@xmath3372 34 14.3&12.22 & 1.85 & 0.92 & 3.531 & -8.82 & 140.4&smc & k7 i & m0 iab + 032188&00 55 03.71&@xmath3373 00 36.6&12.40 & 1.75 & 0.86 & 3.544 & -8.34 & 154.1&smc & k5 i & + 033610&00 55 26.82&@xmath3372 35 56.2&12.60 & 1.75 & 0.91 & 3.531 & -8.44 & 157.4&smc & k7 i & m0 iab + 034158&00 55 36.58&@xmath3372 36 23.6&12.79 & 1.78 & 0.95 & 3.531 & -8.25 & 139.0&smc & k7 i & k5-m0 + 035231&00 55 55.10&@xmath3372 40 30.4&12.02 & 1.32 & 0.66 & 3.568 & -8.34 & 151.8&smc & k2 i & + 037994&00 56 43.55&@xmath3372 30 15.0&12.65 & 1.68 & 0.97 & 3.531 & -8.39 & 148.6&smc & k7 i & k5-m0 + 041778&00 57 56.45&@xmath3372 17 33.3&12.52 & 1.08 & 0.82 & 3.531 & -8.52 & 178.9&smc & k7 i & + 042319&00 58 06.61&@xmath3372 20 59.8&13.09 & 1.90 & 0.94 & 3.556 & -7.41 & 184.9&smc & k2 - 5 i & + 042438&00 58 08.71&@xmath3372 19 26.7&13.20 & 1.59 & 0.87 & 3.552 & -7.35 & 176.5&smc & k3 - 5 i & + 043219&00 58 23.30&@xmath3372 48 40.7&13.06 & 1.84 & 0.94 & 3.518 & -8.35 & 135.7&smc & m0 i & m0 iab + 043725&00 58 33.21&@xmath3372 19 15.6&13.50 & 1.56 & 0.96 & 3.544 & -7.24 & 182.7&smc & k5 i & + 044719&00 58 53.33&@xmath3372 08 35.3&12.98 & 1.53 & 0.82 & & & 95.4&fgd?&k5 v ?
& + 044724&00 58 53.54&@xmath3372 40 38.7&11.78 & 1.59 & 0.87 & & & 55.5&fgd & dwarf & + 044763&00 58 54.44&@xmath3372 41 40.8&12.73 & 1.28 & 0.82 & & & 18.2&fgd & dwarf & + 045378&00 59 07.16&@xmath3372 13 08.6&12.93 & 1.56 & 0.92 & 3.544 & -7.81 & 179.9&smc & k5 i & k5
i + 045850&00 59 16.90&@xmath3372 25 10.9&12.88 & 1.76 & 0.87 & 3.568 & -7.48 & 141.8&smc & k0 - 5 i & k5-m0 + 046497&00 59 31.33&@xmath3372 15 46.4&12.40 & 1.98 & 0.99 & 3.505 & -9.46 & 166.3&smc & m1
i & m0 ia- + 046662&00 59 35.04&@xmath3372 04 06.2&12.90 & 1.88 & 1.07 & 3.491 & -9.54 & 180.2&smc & m2 i & m0 ia + 046910&00 59 40.58&@xmath3372 20 55.9&12.82 & 1.75 & 0.85 & 3.552 & -7.73 & 160.2&smc & k3 - 5
i & m0 ia + 047757&01 00 00.63&@xmath3372 19 40.2&12.52 & 1.87 & 1.02 & 3.505 & -9.34 & 161.1&smc & m1 i & k5-m0 + 048122&01 00 09.42&@xmath3372 08 44.5&12.19 & 1.78 & 0.89 & 3.556 & -8.31 & 172.8&smc & k3 i & + 049033&01 00 30.43&@xmath3371 58 24.7&12.50 & 1.82 & 0.91 & 3.544 & -8.24 & 160.2&smc & k5 i & m0 i + 049428&01 00 40.32&@xmath3372 35 58.8&12.97 & 1.73 & 0.87 & 3.544 & -7.77 & 134.4&smc & k0 - 7 i & k5
i + 049478&01 00 41.56&@xmath3372 10 37.0&12.17 & 1.81 & 0.99 & 3.518 & -9.24 & 177.1&smc & m0 i & k5 ia + 049990&01 00 54.13&@xmath3372 51 36.3&12.20 & 1.66 & 0.85 & 3.544 & -8.54 & 186.8&smc & k5 i & k5 ia + 050237&01 01 00.31&@xmath3372 13 41.6&12.91 & 1.62 & 0.84 & 3.556 & -7.59 & 179.2&smc & k2 - 5 i & k5
i + 050348&01 01 03.26&@xmath3372 04 39.4&12.92 & 1.44 & 0.84 & 3.531 & -8.12 & 179.3&smc & k7 i & + 050360&01 01 03.58&@xmath3372 02 58.5&13.09 & 1.61 & 0.86 & 3.544 & -7.65 & 163.7&smc & k5 i & + 050840&01 01 15.99&@xmath3372 13 10.0&12.57 & 1.95 & 1.02 & 3.499 & -9.55 & 179.9&smc & m1 - 2 i & + 051000&01 01 19.92&@xmath3372 05 13.1&12.89 & 1.66 & 0.85 & 3.544 & -7.85 & 177.7&smc & k5 i & k5
i + 051265&01 01 26.89&@xmath3372 01 41.3&12.87 & 1.51 & 0.86 & 3.552 & -7.68 & 159.1&smc & k3 - 5 i & + 051694&01 01 37.77&@xmath3371 54 16.3&11.83 & 1.19 & 0.72 & & & 17.1&fgd & g v & + 051906&01 01 43.57&@xmath3372 38 25.1&13.02 & 1.29 & 0.83 & 3.544 & -7.72 & 148.1&smc & k5 i & + 052334&01 01 54.16&@xmath3371 52 18.8&12.89 & 1.94 & 0.99 & 3.531 & -8.15 & 165.5&smc & k7 i & m0 iab + 052389&01 01 55.43&@xmath3372 00 29.5&12.85 & 1.60 & 0.91 & 3.531 & -8.19 & 183.8&smc & k7 i & k2
i + 053557&01 02 23.71&@xmath3372 55 21.2&12.72 & 1.77 & 0.91 & 3.531 & -8.32 & 170.8&smc & k7 i & m0
i + 053638&01 02 25.83&@xmath3372 38 56.9&13.16 & 1.83 & 0.89 & 3.544 & -7.58 & 153.4&smc & k2 - 7 i & + 054111&01 02 37.22&@xmath3372 16 25.1&12.55 & 1.74 & 0.87 & 3.568 & -7.81 & 153.6&smc & k0 - 5 i & k5-m0 + 054300&01 02 42.12&@xmath3372 37 29.1&13.02 & 1.74 & 0.89 & 3.568 & -7.34 & 153.8&smc & k0 - 5 i & + 054414&01 02 44.82&@xmath3372 01 51.9&12.93 & 1.65 & 0.85 & 3.552 & -7.62 & 174.0&smc & k3 - 5 i & + 054708&01 02 51.37&@xmath3372 24 15.5&12.82 & 1.81 & 0.91 & 3.540 & -8.01 & 136.8&smc & k0 i & m0 iab + 055188&01 03 02.38&@xmath3372 01 52.9&14.96 & 2.25 & 1.48 & 3.491 & -7.48 & 176.8&smc & m2
i & + 055275&01 03 04.34&@xmath3372 34 12.8&12.91 & 1.70 & 1.02 & 3.525 & -8.31 & 212.2&smc & k7-m0 i&k5-m0 + 055355&01 03 06.43&@xmath3372 28 35.1&12.45 & 1.86 & 0.95 & 3.525 & -8.77 & 137.6&smc & k7-m0 i&k5-m0 + 055462&01 03 08.80&@xmath3372 44 55.1&12.21 & 1.38 & 0.83 & & & 3.2&fgd & dwarf & + 055470&01 03 08.88&@xmath3371 55 50.8&13.12 & 1.75 & 0.86 & 3.552 & -7.43 & 145.1&smc & k3 - 5 i & + 055560&01 03 10.93&@xmath3372 18 32.9&12.96 & 1.66 & 0.90 & 3.552 & -7.59 & 159.2&smc & k3 - 5 i & k5-m0 + 055681&01 03 12.98&@xmath3372 09 26.5&12.52 & 1.65 & 0.96 & 3.478&-10.53 & 182.3&smc & m3 i & m0-m1 + 055933&01 03 18.56&@xmath3372 06 46.2&12.53 & 0.98 & 0.75 & 3.552 & -8.02 & 178.8&smc & k3 - 5 i & + 056389&01 03 27.61&@xmath3372 52 09.4&11.85 & 2.01 & 1.01 & 3.538 & -9.03 & 157.0&smc & k5 - 7 i & m0 i + 056732&01
03 34.30&@xmath3372 06 05.8&12.86 & 1.53 & 0.94 & 3.531 & -8.18 & 183.1&smc & k7 i & + 057386&01 03 47.35&@xmath3372 01 16.0&12.71 & 1.57 & 0.85 & 3.552 & -7.84 & 170.3&smc & k3 - 5 i & k5-m0 + 057472&01 03 48.89&@xmath3372 02 12.7&12.80 & 1.83 & 0.88 & 3.538 & -8.08 & 175.9&smc & k5 - 7 i & k5-m0 + 058100&01 04 01.64&@xmath3372 08 25.2&11.14 & 1.42 & 0.76 & & & -0.3&fgd & k2 - 7 v & + 058149&01 04 02.77&@xmath3372 05 27.7&12.96 & 1.48 & 0.85 & 3.556 & -7.54 & 177.1&smc & k2 - 5 i & k5-m0 + 058472&01 04 09.52&@xmath3372 50 15.3&13.34 & 1.82 & 0.96 & 3.520 & -8.02 & 181.8&smc & k0 i & + 058738&01 04 15.46&@xmath3372 45 19.9&12.75 & 1.60 & 0.76 & 3.568 & -7.61 & 162.1&smc & k2 i & + 058839&01 04 17.71&@xmath3371 57 32.5&13.21 & 1.81 & 0.95 & 3.568 & -7.15 & 192.5&smc & k2 i & + 059426&01 04 30.26&@xmath3372 04 36.1&13.08 & 1.80 & 0.99 & 3.538 & -7.80 & 167.0&smc & k5 - 7 i & k5-m0 + 059803&01 04 38.16&@xmath3372 01 27.2&11.98 & 1.95 & 0.98 & 3.512 & -9.65 & 200.0&smc & m0 - 1 i & + 060447&01 04 53.05&@xmath3372 47 48.5&13.09 & 1.64 & 0.94 & 3.518 & -8.32 & 172.6&smc & m0 i & + 061296&01 05 11.50&@xmath3372 02 27.5&13.07 & 1.77 & 0.92 & 3.531 & -7.97 & 160.1&smc & k7 i & + 062427&01 05 40.04&@xmath3371 58 46.4&13.06 & 1.73 & 0.85 & 3.544 & -7.68 & 170.5&smc & k5 i & + 062763&01 05 50.26&@xmath3371 58 02.2&12.12 & 1.20 & 0.72 & & & 46.3&fgd & k v & + 063114&01 06 01.37&@xmath3372 52 43.2&12.83 & 1.88 & 0.94 & 3.538 & -8.05 & 194.8&smc & k5 - 7 i & + 063131&01
06 01.72&@xmath3372 24 03.8&12.97 & 1.67 & 0.85 & 3.544 & -7.77 & 168.1&smc & k5 i & + 063188&01 06 03.21&@xmath3372 52 16.0&13.07 & 1.85 & 0.92 & 3.568 & -7.29 & 175.3&smc & k2 i & + 064448&01 06 40.21&@xmath3372 28 45.2&12.68 & 1.55 & 0.76 & 3.544 & -8.06 & 155.5&smc & k2 - 7 i & m0 ia- + 064663&01 06 47.62&@xmath3372 16 11.9&11.87 & 1.40 & 0.88 & 3.531 & -9.17 & 139.3&smc & k7 i & + 066066&01 07 29.36&@xmath3372 30 45.7&12.57 & 1.69 & 0.82 & 3.544 & -8.17 & 148.6&smc & k5 i & + 066510&01 07 43.12&@xmath3372 12 15.1&11.69 & 1.21 & 0.66 & & & -60.1&fgd & k5 v & + 066694&01 07 48.88&@xmath3372 23 42.4&12.52 & 1.76 & 0.91 & 3.544 & -8.22 & 137.6&smc & k5 i & + 066754&01 07 50.91&@xmath3372 10 46.8&13.02 & 1.35 & 0.94 & & & -40.1&fgd & g v & + 067509&01 08 13.34&@xmath3372 00
02.9&12.74 & 1.68 & 0.86 & 3.568 & -7.62 & 153.2&smc & k2 i & + 067554&01 08 14.65&@xmath3372 46 40.8&12.64 & 1.62 & 0.84 & 3.538 & -8.24 & 195.9&smc & k5 - 7 i & + 068648&01 08 52.08&@xmath3372 23 07.0&12.33 & 1.76 & 0.87 & 3.568 & -8.03 & 181.4&smc & k2 i & + 069317&01 09 17.09&@xmath3372 12 42.6&12.86 & 1.59 & 1.08 & & & 13.1&fgd & m3 - 4 v & + 070859&01 10 19.89&@xmath3372 03 34.8&11.06 & 1.39 & 0.88 & & & 27.2&fgd & dwarf & + 071507&01 10 50.25&@xmath3372 00 14.5&13.26 & 1.74 & 0.89 & 3.552 & -7.29 & 152.7&smc & k3 - 5 i & + 071566&01 10 53.51&@xmath3372 25 40.0&13.00 & 1.76 & 0.89 & 3.531 & -8.04 & 188.8&smc & k7 i & + 081668&01 24 54.03&@xmath3373 26 49.2&13.14 & 1.77 & 0.90 & 3.544 & -7.60 & 161.3&smc & & + 081961&01 25 38.80&@xmath3373 21 55.6&11.84 & 1.90 & 0.94 & 3.528 & -9.28 & 160.2&smc & & + 082159&01 26 09.91&@xmath3373 23 15.4&12.71 & 1.71 & 0.83 & 3.573 & -7.60 & 167.2&smc & & + 083202&01 29 18.52&@xmath3373 01 59.3&11.53 & 1.08 & 0.82 & 3.577 & -8.73 & 158.6&smc & & + 083593&01 30 33.92&@xmath3373 18 41.9&12.64 & 1.87 & 1.00 & 3.491 & -9.80 & 180.1&smc & & m2 ia + 084202&01 33 08.98&@xmath3373 25 32.5&12.00 & 1.33 & 0.70 & & & 95.3&fgd ? & & + 084392&01 34 08.70&@xmath3373 06 04.5&11.48 & 1.55 & 1.33 & & & -21.4&fgd & & + l c c c c c c r r l l l 009392&04 50 58.71&@xmath3369 14 03.2&12.88 & 1.95 & 1.07 & 3.533 & -7.99 & 263.9&lmc & & + 010895&04 51 30.99&@xmath3369 14 52.0&13.55 & 1.94 & 1.23 & 3.479 & -9.31 & 263.2&lmc & & + 011656&04 51 47.29&@xmath3369 19 25.1&13.01 & 1.89 & 1.01 & 3.553 & -7.39 & 276.9&lmc & & + 016332&04 53 14.78&@xmath3369 12 18.3&13.73 & 2.17 & 1.39 & 3.477 & -9.22 & 276.8&lmc & & + 016554&04 53 18.50&@xmath3369 17 03.5&12.88 & 1.20 & 0.97 & 3.566 & -7.37 & 267.8&lmc & & + 017261&04 53 30.84&@xmath3369 17 49.8&13.00 & 1.03 & 1.20 & 3.489 & -9.39 & 262.1&lmc & & + 017338&04 53 32.34&@xmath3369 01 17.8&10.93 & 1.54 & 0.85 & & & -13.9&fgd & & + 018456&04 53 51.19&@xmath3368 50 04.3&12.57 & 1.60 & 0.98 & & & 83.0&fgd & & + 021369&04 54 36.90&@xmath3369 20 22.7&11.26 & 1.77 & 0.87 & 3.600 & -8.63 & 256.1&lmc & & + 021480&04 54 38.56&@xmath3369 11 17.4&13.19 & 1.82 & 1.48 & 3.477 & -9.76 & 256.1&lmc & & + 021534&04 54 39.46&@xmath3369 04 36.7&12.63 & 2.02 & 1.07 & 3.533 & -8.24 & 270.4&lmc & & + 022204&04 54 49.77&@xmath3369 30 03.0&12.72 & 1.98 & 0.99 & 3.560 & -7.60 & 261.7&lmc & & + 023095&04 55 03.09&@xmath3369 29 13.4&14.38 & 2.00 & 1.79 & 3.477 & -8.57 & 252.0&lmc & & + 024014&04 55 16.11&@xmath3369 19 12.8&12.82 & 1.47 & 1.23 & 3.479&-10.04 & 246.5&lmc & & + 024410&04 55 21.72&@xmath3369 47 17.2&14.45 & 2.07 & 1.57 & 3.477 & -8.50 & 257.1&lmc & & + 024987&04 55 30.05&@xmath3369 29 11.1&12.08 & 2.04 & 1.09 & 3.526 & -8.97 & 260.7&lmc & & + 025818&04 55 41.86&@xmath3369 26 24.8&11.72 & 2.06 & 1.05 & 3.539 & -8.99 & 253.6&lmc & & + 026286&04 55 48.28&@xmath3369 24 07.1&12.37 & 1.96 & 1.04 & 3.543 & -8.27 & 256.1&lmc & & + 028780&04 56 23.70&@xmath3369 42 11.9&12.76 & 1.83 & 0.96 & 3.570 & -7.45 & 262.8&lmc & & + 029153&04 56 28.30&@xmath3369 40 37.6&12.85 & 1.79 & 0.98 & 3.563 & -7.44 & 268.6&lmc & & + 030861&04 56 48.61&@xmath3369 39 55.9&12.25 & 1.92 & 1.09 & 3.526 & -8.80 & 249.9&lmc & & + 030929&04 56 49.63&@xmath3369 48 32.0&12.06 & 1.64 & 0.73 & 3.647 & -7.43 & 248.8&lmc & & + 035415&04 57 44.66&@xmath3369 30 35.0&13.37 & 1.93 & 1.10 & 3.523 & -7.78 & 260.9&lmc & & + 038347&04 58 21.08&@xmath3369 33 38.3&11.30 & 1.35 & 0.73 & & & 5.1&fgd & & + 054365&05 02 09.57&@xmath3370 25 02.4&13.26 & 1.85 & 1.10 & 3.506 & -8.45 & 237.0&lmc & m3 i & + 058820&05 03 15.36&@xmath3370 17 41.9&13.25 & 1.81 & 1.03 & 3.544 & -7.37 & 251.8&lmc & m0 i & + 062090&05 04 05.10&@xmath3370 22 46.7&12.50 & 1.96 & 1.00 & 3.531 & -8.40 & 243.8&lmc & m1 i & + 062353&05 04 09.92&@xmath3370 12 18.0&12.86 & 1.49 & 1.00 & 3.531 & -8.04 & 237.2&lmc & m1 i & + 064048&05 04 41.79&@xmath3370 42 37.2&13.28 & 1.89 & 1.19 & 3.506 & -8.43 & 240.4&lmc & m3 i & + 064706&05 04 54.13&@xmath3370 33 18.9&12.79 & 1.63 & 0.98 & 3.538 & -7.96 & 238.3&lmc & m0 - 1 i & + 065558&05 05 10.03&@xmath3370 40 03.2&12.62 & 1.89 & 1.01 & 3.544 & -8.00 & 246.3&lmc & m0 i & + 066778&05 05 33.44&@xmath3370 33 47.1&12.92 & 1.40 & 1.19 & 3.484 & -9.70 & 240.6&lmc & m4 i & + 067982&05 05 56.61&@xmath3370 35 24.0&12.76 & 1.93 & 1.09 & 3.477&-10.19 & 244.1&lmc & m4.5 i & + 068098&05 05 58.92&@xmath3370 29 14.6&13.11 & 1.90 & 1.04 & 3.531 & -7.79 & 240.2&lmc & m1 i & + 068125&05 05 59.56&@xmath3370 48 11.4&13.43 & 1.83 & 1.20 & 3.484 & -9.19 & 224.3&lmc & m4 i & m5iab + 069960&05 06 36.42&@xmath3370 32 38.7&13.10 & 1.91 & 1.02 & 3.518 & -8.18 & 242.3&lmc & m2 i & + 071357&05 07 05.62&@xmath3370 32 44.3&11.70 & 2.07 & 1.09 & 3.531 & -9.20 & 241.8&lmc & m1 i & + 072727&05 07 32.52&@xmath3370 39 04.6&13.08 & 2.15 & 1.20 & 3.518 & -8.20 & 234.4&lmc & m2 i & + 106201&05 17 09.11&@xmath3369 32 21.1&13.29 & 1.51 & 1.24 & 3.518 & -7.99 & 261.0&lmc & m2 i & + 109106&05 17 56.51&@xmath3369 40 25.4&12.96 & 1.85 & 1.02 & 3.518 & -8.32 & 248.8&lmc & m2 i & + 113364&05 19 03.35&@xmath3369 39 55.2&11.70 & 1.46 & 0.93 & 3.531 & -9.20 & 253.0&lmc & m1 i & + 116895&05 19 53.34&@xmath3369 27 33.4&12.43 & 1.92 & 1.03 & 3.506 & -9.28 & 264.3&lmc & m3 i & + 119219&05 20 23.69&@xmath3369 33 27.3&12.14 & 2.04 & 0.98 & 3.506 & -9.57 & 259.4&lmc & m3 i & + 123778&05 21 28.06&@xmath3369 30 16.5&13.49 & 1.78 & 1.10 & 3.506 & -8.22 & 274.2&lmc & m3 i & + 124836&05 21 43.54&@xmath3369 21 27.6&13.19 & 1.59 & 1.01 & 3.553 & -7.21 & 274.6&lmc & & + 126683&05 22 11.01&@xmath3369 17 24.2&11.60 & 1.24 & 0.67 & & & 93.3&fgd & k2 v & + 128130&05 22 31.21&@xmath3369 34 05.1&13.07 & 1.86 & 1.00 & 3.518 & -8.21 & 259.0&lmc & m2 i & + 130426&05 23 02.84&@xmath3369 20 37.1&13.18 & 1.88 & 1.05 & 3.531 & -7.72 & 256.4&lmc & m1 i & + 131735&05 23 34.09&@xmath3369 19 07.0&12.65 & 1.84 & 0.89 & 3.556 & -7.71 & 234.8&lmc & k7 i & + 134383&05 25 44.95&@xmath3369 04 48.9&13.46 & 1.65 & 1.21 & 3.506 & -8.25 & 268.4&lmc & m3 i & m3 i + 135720&05 26 27.52&@xmath3369 10 55.5&13.57 & 1.85 & 1.35 & 3.506 & -8.14 & 269.9&lmc & m3 i & + 135754&05 26 28.32&@xmath3369 07 57.4&13.07 & 1.96 & 1.05 & 3.531 & -7.83 & 279.0&lmc & m1 i & + 136042&05 26 34.92&@xmath3368 51 40.1&12.24 & 1.08 & 1.09 & 3.531 & -8.66 & 266.0&lmc & m1 i & m2 i + + 136348&05 26 42.20&@xmath3368 56 38.7&13.11 & 1.89 & 1.05 & 3.531 & -7.79 & 276.7&lmc & m1 i & + 136378&05 26 42.79&@xmath3368 57 13.4&13.28 & 1.97 & 1.11 & 3.506 & -8.43 & 301.0&lmc & m3 i & + 137624&05 27 10.38&@xmath3369 16 17.6&13.16 & 1.88 & 1.02 & 3.544 & -7.46 & 279.1&lmc & m0 i & + 137818&05 27 14.33&@xmath3369 11 10.7&13.33 & 1.74 & 1.20 & 3.506 & -8.38 & 274.9&lmc & m3 i & + 138405&05 27 26.86&@xmath3369 00 02.0&13.08 & 1.83 & 1.02 & 3.544 & -7.54 & 271.8&lmc & m0 i & m0 iab + 138475&05 27 28.16&@xmath3369 00 36.0&12.65 & 1.66 & 1.03 & 3.544 & -7.97 & 271.0&lmc & m0 i & m1 ia- + 138552&05 27 29.84&@xmath3367 14 12.9&12.80 & 1.54 & 1.18 & 3.531 & -8.10 & 296.1&lmc & & m1 ia + 139027&05 27 39.72&@xmath3369 09 01.1&12.13 & 1.15 & 0.92 & 3.556 & -8.23 & 281.4&lmc & k7 i & m1 ia + 139413&05 27 47.62&@xmath3369 13 20.3&12.68 & 1.53 & 1.17 & 3.506 & -9.03 & 272.8&lmc & m3 i & + 139588&05 27 51.22&@xmath3367 18 04.3&13.19 & 1.83 & 1.05 & 3.531 & -7.71 & 292.9&lmc & m1 i & + 139591&05 27 51.28&@xmath3369 10 45.8&12.54 & 1.40 & 0.96 & 3.525 & -8.53 & 265.5&lmc & m1 - 2 i & + 140006&05 28 00.12&@xmath3369 07 42.3&13.05 & 1.71 & 0.97 & 3.512 & -8.44 & 267.8&lmc & m2 - 3 i & m0 ia + 140296&05 28 06.11&@xmath3369 07 13.5&13.12 & 1.87 & 1.18 & 3.525 & -7.95 & 271.2&lmc & m1 - 2 i & m0 ia + 140403&05 28 08.18&@xmath3369 13 10.8&13.01 & 2.01 & 1.15 & 3.484 & -9.61 & 267.7&lmc & m3 - 5 i & + 140782&05 28 16.01&@xmath3369 12 01.1&13.00 & 1.67 & 1.03 & 3.531 & -7.90 & 271.4&lmc & m1 i & + 140912&05 28 18.69&@xmath3369 07 34.7&12.83 & 1.13 & 0.97 & 3.531 & -8.07 & 276.4&lmc & m1 i & m1 ia- + 141377&05 28 28.01&@xmath3369 12 57.2&10.93 & 1.61 & 0.70 & 3.657 & -8.48 & 272.1&lmc & k0 i & + 141507&05 28 30.42&@xmath3369 00 44.7&12.98 & 1.90 & 1.01 & 3.544 & -7.64 & 285.1&lmc & m0 i & + 141568&05 28 31.63&@xmath3369 05 31.2&13.23 & 2.02 & 1.19 & 3.512 & -8.26 & 272.5&lmc & m2 - 3 i & m2 iab + 142102&05 28 43.26&@xmath3367 18 28.5&12.84 & 1.89 & 0.99 & 3.556 & -7.52 & 311.8&lmc & k7 i & + 142202&05 28 45.59&@xmath3368 58 02.3&12.15 & 1.65 & 1.03 & 3.538 & -8.60 & 272.3&lmc & m0-m1 i&m0-m1
+ 142907&05 29 00.86&@xmath3368 46 33.6&13.05 & 1.89 & 1.06 & 3.531 & -7.85 & 273.1&lmc & m1 i & + 143035&05 29 03.58&@xmath3369 06 46.3&13.52 & 1.93 & 1.27 & 3.484 & -9.10 & 268.3&lmc & m3 - 4.5 & + 143137&05 29 05.59&@xmath3367 18 18.0&12.79 & 1.11 & 0.91 & 3.544 & -7.83 & 314.5&lmc & m0 i & m0 iab + 143280&05 29 08.49&@xmath3369 12 18.6&13.27 & 1.94 & 1.13 & 3.506 & -8.44 & 269.3&lmc & m3 i & + 143877&05 29 21.10&@xmath3368 47 31.5&11.82 & 1.94 & 0.95 & 3.556 & -8.54 & 273.2&lmc & k7 i & m1ia + 143898&05 29 21.49&@xmath3369 00 20.3&11.96 & 0.55 & 0.76 & 3.525 & -9.11 & 285.5&lmc & m1 - 2 i & + 144217&05 29 27.66&@xmath3369 08 50.3&12.23 & 1.67 & 1.13 & 3.531 & -8.67 & 267.9&lmc & m1 i & m1 ia + 145013&05 29 42.32&@xmath3368 57 17.3&12.15 & 1.89 & 1.16 & 3.518 & -9.13 & 273.0&lmc & m2 i & m1ia + 145112&05 29 44.02&@xmath3369 05 50.2&12.31 & 2.08 & 1.07 & 3.512 & -9.18 & 263.5&lmc & m2 - 3 & + 145716&05 29 54.85&@xmath3369 04 15.6&12.49 & 1.86 & 0.96 & 3.538 & -8.26 & 285.0&lmc & m0 - 1 & + 145728&05 29 55.04&@xmath3367 18 36.9&12.45 & 1.19 & 1.02 & 3.506 & -9.26 & 308.2&lmc & m3 i & m1ia
+ + 146126&05 30 02.36&@xmath3367 02 45.0&11.17 & 1.80 & 0.84 & 3.568 & -9.05 & 314.3&lmc & k5 i & + 146244&05 30 04.63&@xmath3368 47 28.9&12.92 & 1.92 & 0.98 & 3.556 & -7.44 & 275.0&lmc & k7 i & m0 iab + 146266&05 30 04.99&@xmath3369 03 59.9&13.15 & 1.84 & 1.01 & 3.556 & -7.21 & 269.2&lmc & k7 i & + 146548&05 30 09.67&@xmath3369 11 03.9&13.80 & 2.08 & 1.18 & 3.550 & -6.70 & 277.2&lmc & k7-m0 i & + 147199&05 30 21.00&@xmath3367 20 05.7&12.73 & 1.57 & 1.20 & 3.484 & -9.89 & 309.4&lmc & m4 i & m1 ia + 147257&05 30 22.20&@xmath3367 06 31.4&12.76 & 1.45 & 0.94 & 3.531 & -8.14 & 302.1&lmc & m1 i & + 147276&05 30 22.49&@xmath3367 05 05.9&11.94 & 1.33 & 0.71 & & & 63.3&fgd & k2 v & + 147372&05 30 24.36&@xmath3367 29 13.0&11.91 & 1.31 & 0.73 & & & 56.1&fgd & k2 v & + 147479&05 30 26.37&@xmath3369 30 24.7&12.78 & 1.90 & 1.02 & 3.506 & -8.93 & 266.6&lmc & m3 i & + 147928&05 30 33.55&@xmath3367 17 15.4&12.38 & 1.30 & 0.95 & 3.556 & -7.98 & 291.1&lmc & k7 i & m2 i + + 148035&05 30 35.61&@xmath3368 59 23.6&13.88 & 1.66 & 1.38 & 3.484 & -8.74 & 284.5&lmc & m4 i & + 148041&05 30 35.69&@xmath3367 12 04.3&13.06 & 1.81 & 0.99 & 3.531 & -7.84 & 308.2&lmc & m1 i & + 148381&05 30 41.58&@xmath3369 15 33.7&12.24 & 1.86 & 1.10 & 3.477&-10.71 & 272.6&lmc & m4.5 - 5 & + 148409&05 30 42.10&@xmath3369 05 23.2&13.32 & 1.81 & 1.05 & 3.538 & -7.43 & 268.9&lmc & m0 - 1 i & m1 iab + 148600&05 30 45.25&@xmath3367 07 59.2&13.23 & 1.90 & 1.11 & 3.506 & -8.48 & 305.2&lmc & m3 i & + 149026&05 30 52.38&@xmath3367 17 34.5&12.80 & 1.43 & 0.93 & 3.531 & -8.10 & 307.6&lmc & m1 i & m2 i + + 149065&05 30 53.17&@xmath3367 30 52.0&12.09 & 1.21 & 0.68 & & & -7.9&fgd & k0 v & + 149560&05 31 00.62&@xmath3369 10 39.6&13.05 & 0.60 & 1.20 & 3.489 & -9.34 & 279.9&lmc & comp i & + 149587&05 31 01.19&@xmath3369 10 59.2&12.51 & 0.69 & 0.82 & 3.544 & -8.11 & 278.1&lmc & m0 i & + 149721&05 31 03.50&@xmath3369 05 40.0&12.71 & 1.86 & 0.97 & 3.562 & -7.58 & 277.3&lmc & k5 - 7 i & m1 iab + 149767&05 31 04.33&@xmath3369 19 02.9&13.10 & 2.03 & 1.29 & 3.506 & -8.61 & 274.4&lmc & m3 i & + 150040&05 31 09.35&@xmath3367 25 55.1&12.81 & 1.96 & 1.20 & 3.484 & -9.81 & 280.0&lmc & m4 i & m4 ia- + 150396&05 31 15.58&@xmath3369 03 58.8&13.26 & 1.81 & 1.15 & 3.525 & -7.81 & 271.5&lmc & m1 - 2 i & + 150577&05 31 18.56&@xmath3369 09 28.2&13.27 & 1.80 & 1.08 & 3.562 & -7.02 & 277.8&lmc & k5 - 7 i & + 150976&05 31 25.82&@xmath3369 21 17.9&13.17 & 1.85 & 1.06 & 3.531 & -7.73 & 275.9&lmc & m1 i & + 152132&05 31 47.50&@xmath3367 23 03.3&13.16 & 1.90 & 1.06 & 3.544 & -7.46 & 292.5&lmc & m0 i & m0 ia- + 153298&05 32 08.91&@xmath3367 11 18.6&13.11 & 1.85 & 1.03 & 3.531 & -7.79 & 300.9&lmc & m1 i & + 153866&05 32 19.30&@xmath3367 25 00.5&13.16 & 1.82 & 1.01 & 3.531 & -7.74 & 297.2&lmc & m1 i & + 154311&05 32 27.54&@xmath3369 16 53.0&12.56 & 1.89 & 1.06 & 3.506 & -9.15 & 261.0&lmc & m3 i & + 154542&05 32 31.52&@xmath3369 20 25.7&13.02 & 1.94 & 1.01 & 3.544 & -7.60 & 272.6&lmc & m0 i & + 154729&05 32 35.44&@xmath3369 07 51.9&13.21 & 1.51 & 1.08 & 3.525 & -7.86 & 292.0&lmc & m1 - 2 i & + 155529&05 32 50.32&@xmath3367 27 45.3&13.34 & 1.84 & 1.20 & 3.506 & -8.37 & 292.3&lmc & m3 i & + 156794&05 33 14.53&@xmath3367 03 48.5&12.95 & 1.79 & 1.04 & 3.531 & -7.95 & 302.7&lmc & m1 i & + 157401&05 33 26.88&@xmath3367 04 13.7&12.27 & 1.99 & 1.06 & 3.506 & -9.44 & 299.1&lmc & m3 i & + 157533&05 33 29.67&@xmath3367 31 38.0&13.16 & 1.50 & 0.99 & 3.568 & -7.06 & 302.2&lmc & k5 i & m1 ia + 158317&05 33 44.60&@xmath3367 24 16.9&13.35 & 1.96 & 1.12 & 3.518 & -7.93 & 301.2&lmc & m2 i & + 158646&05 33 52.26&@xmath3369 11 13.2&13.10 & 2.23 & 1.33 & 3.496 & -8.97 & 288.3&lmc & m3 - 4 i & + 159893&05 34 19.57&@xmath3368 59 36.4&13.10 & 1.96 & 1.06 & 3.536 & -7.69 & 289.9&lmc & & + 159974&05 34 21.49&@xmath3369 21 59.8&12.72 & 1.77 & 0.91 & 3.580 & -7.38 & 250.7&lmc & k2 - 5 i & + 160170&05 34 25.97&@xmath3369 21 47.7&11.03 & 1.53 & 0.82 & & & 42.3&fgd & k2 v & + 160518&05 34 33.90&@xmath3369 15 02.3&13.10 & 1.89 & 1.17 & 3.525 & -7.97 & 300.7&lmc & m1 - 2 i & + 161078&05 34 47.07&@xmath3369 29 00.1&12.91 & 1.61 & 1.02 & 3.544 & -7.71 & 269.4&lmc & m0 i & + 162635&05 35 24.61&@xmath3369 04 03.2&14.23 & 2.33 & 1.36 & 3.531 & -6.67 & 296.9&lmc & m1 i & + 163007&05 35 32.84&@xmath3369 04 18.6&13.07 & 1.51 & 1.04 & 3.556 & -7.29 & 293.8&lmc & k7 i & + 163466&05 35 43.86&@xmath3368 51 21.1&12.45 & 1.76 & 1.05 & 3.539 & -8.26 & 284.6&lmc & & + 163814&05
35 52.01&@xmath3369 22 28.5&12.75 & 1.80 & 0.99 & 3.544 & -7.87 & 267.1&lmc & m0 i & + 164506&05 36 06.44&@xmath3368 56 40.8&12.87 & 1.44 & 0.97 & 3.566 & -7.38 & 288.0&lmc & & + 164709&05 36 10.56&@xmath3368 54 40.5&12.02 & 0.40 & 0.67 & 3.667 & -7.33 & 287.8&lmc & & + 165242&05 36 20.42&@xmath3368 56 18.9&13.97 & 1.95 & 1.28 & 3.477 & -8.98 & 289.2&lmc & & + 165543&05 36 26.91&@xmath3369 23 50.7&10.98 & 1.62 & 0.81 & 3.620 & -8.73 & 277.1&lmc & k0 i & + 166155&05 36 40.60&@xmath3369 23 16.4&12.94 & 1.51 & 0.96 & 3.556 & -7.42 & 263.5&lmc & k7 i & + 168047&05 37 20.65&@xmath3369 19 38.2&12.47 & 1.54 & 0.97 & 3.568 & -7.75 & 262.1&lmc & k2 - 7 i & + 168290&05 37 26.37&@xmath3368 47 40.1&13.23 & 2.03 & 1.18 & 3.496 & -8.87 & 305.3&lmc & & + 168469&05 37 30.70&@xmath3369 02 33.2&13.50 & 2.24 & 1.14 & 3.562 & -6.79 & 265.8&lmc & k5 - 7 i & + 168757&05 37 36.96&@xmath3369 29 23.5&14.08 & 1.77 & 1.34 & 3.506 & -7.63 & 272.3&lmc & m3 i & + 169049&05 37 43.16&@xmath3369 24 59.6&12.65 & 2.02 & 1.14 & 3.518 & -8.63 & 264.5&lmc & m1 - 3 i & + 169142&05 37 45.15&@xmath3369 20 48.2&12.11 & 0.91 & 0.91 & 3.496 & -9.96 & 264.5&lmc & m3 - 4 i & + 169754&05 37 58.77&@xmath3369 14 23.7&13.21 & 2.15 & 1.13 & 3.591 & -6.77 & 275.2&lmc & k2 - 3 i & + 170079&05 38 06.71&@xmath3369 17 29.5&14.60 & 2.30 & 1.60 & 3.506 & -7.11 & 256.5&lmc & m3 i & + 170452&05 38 16.10&@xmath3369 10 10.9&13.99 & 2.39 & 1.50 & 3.477 & -8.96 & 289.4&lmc & m4.5 - 5 & + 170455&05 38 16.20&@xmath3369 23 31.7&12.08 & 1.36 & 0.90 & & & -29.1&fgd & dwarf & + 170539&05 38 18.24&@xmath3369 17 42.1&13.86 & 2.14 & 1.29 & 3.506 & -7.85 & 261.1&lmc & m3 i & + 173854&05 39 46.25&@xmath3369 19 28.1&13.60 & 2.08 & 1.19 & 3.531 & -7.30 & 244.8&lmc & m1 i & + 174324&05 40 07.72&@xmath3369 20 05.1&13.83 & 1.91 & 1.26 & 3.512 & -7.66 & 255.3&lmc & m2 - 3 i & + 174543&05 40 17.13&@xmath3369 27 53.7&12.97 & 1.58 & 1.06 & 3.506 & -8.74 & 246.9&lmc & m3 i & + 174714&05 40 24.48&@xmath3369 21 16.6&13.13 & 1.98 & 1.21 & 3.477 & -9.82 & 251.0&lmc & m4 - 5 i & + 174742&05 40 25.38&@xmath3369 15 30.2&12.50 & 1.63 & 0.92 & 3.550 & -8.00 & 251.4&lmc & k7-m0 i & + 175015&05 40 37.04&@xmath3369 26 20.1&13.31 & 1.92 & 1.15 & 3.506 & -8.40 & 249.6&lmc & m3 i & + 175188&05 40 43.80&@xmath3369 21 57.8&13.52 & 1.72 & 1.36 & 3.512 & -7.97 & 260.4&lmc & m2 - 3 i & + 175464&05 40 55.36&@xmath3369 23 25.0&12.90 & 2.20 & 1.22 & 3.512 & -8.59 & 245.6&lmc & m2 - 3 i & + 175549&05 40 59.25&@xmath3369 18 36.2&13.24 & 2.23 & 1.39 & 3.512 & -8.25 & 243.9&lmc & m2 - 3 i & m2 i + 175709&05 41 05.17&@xmath3369 04 42.5&12.74 & 1.95 & 1.06 & 3.544 & -7.88 & 251.7&lmc & m0 i & + 175746&05 41 06.94&@xmath3369 17 14.8&13.30 & 2.06 & 1.26 & 3.506 & -8.41 & 262.2&lmc & m3 i & m1 ia- + 176135&05 41 21.89&@xmath3369 31 48.8&13.06 & 2.10 & 1.26 & 3.506 & -8.65 & 255.2&lmc & m3 i & + 176216&05 41 24.60&@xmath3369 18 12.8&13.66 & 1.67 & 1.22 & 3.556 & -6.70 & 257.5&lmc & k7 i & m1 ia- + 176335&05 41 29.70&@xmath3369 27 16.2&12.90 & 2.01 & 1.03 & 3.556 & -7.46 & 247.4&lmc & k7 i & + 176695&05 41 43.49&@xmath3369 28 15.4&12.92 & 1.97 & 1.03 & 3.544 & -7.70 & 248.9&lmc & m0 i & + 176715&05 41 44.05&@xmath3369 12 02.7&13.05 & 1.13 & 0.98 & 3.544 & -7.57 & 243.7&lmc & m0 i & m1 i + 176890&05 41 50.26&@xmath3369 21 15.7&12.85 & 1.97 & 1.01 & 3.556 & -7.51 & 254.8&lmc & k7 i & m0 iab + 177150&05 42 00.84&@xmath3369 11 37.0&13.80 & 1.89 & 1.20 & 3.531 & -7.10 & 249.0&lmc & m1 i & m1 iab + 178066&05 42 38.71&@xmath3369 09 51.4&13.30 & 2.00 & 1.05 & 3.556 & -7.06 & 245.2&lmc & k7 i & m2 ia + 178555&05 43 02.16&@xmath3369 05 49.6&13.04 & 1.97 & 1.09 & 3.544 & -7.58 & 269.6&lmc & m0 i & + l c c c c c k2 i & 4300 & & 3970 & 4000 & -0.97 + k5 i & 4000 & & 3520 & 3800 & -1.20 + k7 i & 3750 & & 3490&3700 & -1.36 + m0 i & 3550&3600 & 3460&3600 & -1.50 + m1 i & 3450&3550 & 3435 & 3500 & -1.71 + m2 i & 3350&3450 & 3340 & 3400 & -2.00 + m3 i & 3250&3200 & 3275 & 3300 & -2.37 + m4 i & 3000&2950 & 3195 & 3150 & -3.09 + m5 i & 2800&2800 & 3070 & 3000 & -4.04 + c c c c c c c
c k2 i & & & & & @xmath76 & @xmath77 & 7 + k5 i & & & & & @xmath78 & @xmath79 & 12 + k7 i & @xmath80 & @xmath81 & 11&&@xmath82 & @xmath83 & 23 + m0 i & @xmath84 & @xmath56 & 14&&@xmath85 & @xmath56 & 4 + m1 i & @xmath86 & @xmath87 & 20 & & & & + m2 i & @xmath88 & @xmath89 & 5 & & & & + m3 i & @xmath90 & @xmath91 & 23 & & & & + m4 i & @xmath92 & @xmath93 & 5 & & & & + l c l l c c + 003196 & 13.11 & skkm13 & m1 i & 3.505 & -8.75 + 018136 & 11.98 & skkm63 & m0 ia & 3.518 & -9.43 + 021362 & 12.89 & skkm78 & k5-m0 i & 3.531 & -8.15 + 021381 & 12.81 & skkm79 & k5-m0 i & 3.531 & -8.23 + 023401 & 12.99 & skkm89 & k5 i & 3.544 & -7.75 + 035445 & 12.74 & skkm144 & m0 iab & 3.518 & -8.67 + 069886 & 11.74 & skkm319 & m2 ia & 3.491 & -10.70 + + 101 - 6 & 12.67 & skkm13 & m1 i & 3.505 & -9.19 + 106 - 1a & 12.24 & skkm63 & m0 ia & 3.518 & -9.17 + 105 - 7 & 12.80 & skkm78 & k5-m0 i & 3.531 & -8.24 + 106 - 5 & 12.95 & skkm79 & k5-m0 i & 3.531 & -8.09 + 106 - 7 & 13.12 & skkm89 & k5 i & 3.544 & -7.62 + 106 - 9 & 13.16 & skkm91 & k5-m0 i & 3.531 & -7.88 + 108 - 3 & 12.56 & skkm110 & m0 i & 3.518 & -8.85 + 105 - 11 & 12.38 & skkm114 & m0 iab & 3.518 & -9.03 + 108 - 8 & 13.19 & skkm129 & k0 - 2 i & 3.568 & -7.17 + 105 - 21 & 13.68 & skkm135 & k5-m0 i & 3.531 & -7.36 + hv838 & 13.35 & skkm142 & m0e i & 3.518 & -8.06 + 114 - 3 & 12.89 & skkm144 & m0 iab & 3.518 & -8.52 + hv11423&11.77 & skkm205 & m0 ia & 3.518 & -9.64 + 115 - 6 & 12.92 & skkm210 & k1 - 3 iab & 3.568 & -7.44 + 116 - 15 & 12.05 & skkm236 & m0 ia & 3.518 & -9.36 + 115 - 17 & 13.03 & skkm237 & k5-m0 iab & 3.531 & -8.01 + 120 - 14 & 11.96 & skkm275 & k5-m0 iab & 3.531 & -9.08 + hv2084 & 12.62 & skkm319 & m2 ia & 3.491 & -9.82 + hv2228 & 12.89 & skkm347 & m0 iab & 3.518 & -8.52 + 108 - 2 & 12.28 & & m0 ia & 3.518 & -9.13 + 118 - 18 & 13.32 & skkm272?&m0 ia & 3.518 & -8.09 + l c l l c c + 141430 & 12.30 & 46 - 32 & m0 ia & 3.544 & -8.32 + 141772 & 12.55 & 46 - 34 & m2 ia & 3.518 & -8.73 + 156011 & 12.11 & 53 - 3 & m0 ia & 3.544 & -8.51 + + 46 - 40 & 12.98 & & m1 ia & 3.531 & -7.92 + 45 - 48 & 13.38 & & m4 ia - iab & 3.484 & -9.24 + 54 - 35 & 12.85 & & m1 i + b & 3.531 & -8.05 + 54 - 47a & 13.10 & & m1 iab & 3.531 & -7.80 + 45 - 2 & 12.90 & & m2 iab & 3.518 & -8.38 + 37 - 32 & 12.95 & & m2 i & 3.518 & -8.33 + 37 - 35 & 12.89 & hv916 & m3 ia & 3.506 & -8.82 + 37 - 24 & 13.59 & hv2360 & m2 ia & 3.518 & -7.69 + 39 - 33 & 12.57 & hv888 & m4 ia & 3.484 & -10.05 + 46 - 2 & 12.25 & hv2450 & m2 ia & 3.518 & -9.03 + 61 - 23 & 13.28 & & m1 ia - iab & 3.531 & -7.62 + 53 - 3 & 12.04 & & m0 ia & 3.544 & -8.58 + 46 - 31 & 13.00 & hv2567 & m2 iab & 3.518 & -8.28 + 46 - 51 & 12.84 & hv2602 & m2 ia - iab & 3.518 & -8.44 + 46 - 52 & 13.40 & & m1 iab & 3.531 & -7.50 + 54 - 47 & 13.02 & & m0 iab & 3.544 & -7.60 + 54 - 39 & 12.76 & hv2781 & m1 ia - iab & 3.531 & -8.14 + 54 - 38 & 13.03 & & m2 ia & 3.518 & -8.25 + 54 - 56 & 13.32 & & m0 ia - iab & 3.544 & -7.30 + 54 - 44 & 13.03 & & m1 ia - iab & 3.531 & -7.87 + 55 - 20 & 13.09 & & m2 ia - iab & 3.518 & -8.19 + 52 - 4 & 13.00 & hv5914 & m1 iab & 3.531 & -7.90 + | we investigate the red supergiant ( rsg ) content of the smc and lmc using multi - object spectroscopy on a sample of red stars previously identified by _ bvr _ ccd photometry .
we obtained high accuracy ( @xmath0 km s@xmath1 ) radial velocities for 118 red stars seen towards the smc and 167 red stars seen towards the lmc , confirming most of these ( 89% and 95% , respectively ) as red supergiants ( rsgs ) .
spectral types were also determined for most of these rsgs .
we find that the distribution of spectral types is skewed towards earlier type at lower metallicities : the average ( median ) spectral type is k5 - 7 i in the smc , m1 i in the lmc , and m2 i in the milky way .
our examination of the kurucz atlas 9 model atmospheres suggests that the effect that metallicity has on the appearance on the tio lines is probably sufficient to account for this effect , and we argue that rsgs in the magellanic clouds are 100@xmath2k ( lmc ) and 300@xmath2k ( smc ) cooler than galactic stars of the same spectral types .
the colors of the kurucz models are not consistent with this interpretation for the smc , although other models ( e.g. , bessel et al . ) show good agreement .
a finer grid of higher - resolution synthetic spectra appropriate to cool supergiants is needed to better determine the effective temperature scale .
we compare the distribution of rsgs in the h - r diagram to that of various stellar evolutionary models ; we find that none of the models produce rsgs as cool and luminous as what is actually observed .
this result is much larger than any uncertainty in the effective temperature scale .
we note that were we to simply adopt the uncorrected galactic effective scale for rsgs and apply this to our sample , then the smc s rsgs would be under luminous compared to the lmc s , contrary to what we expect from stellar evolution considerations . in all of our h - r diagrams
, however , there is an elegant sequence of decreasing effective temperatures with increasing luminosities ; explaining this will be an important test of future stellar evolutionary models .
finally , we compute the blue - to - red supergiant ratio in the smc and lmc , finding that the values are indistinguishable ( @xmath3 ) for the two clouds .
we emphasize that observed " b / r values must be carefully determined if a comparison to that predicted by stellar models is to be meaningful .
the non - rotation geneva models overestimate the number of blue - to - red supergiants for the smc , but underestimate it for the lmc ; however , given the inability to produce high luminosity rsgs in the models that match what is observed in the h - r diagram , such a disagreement is not surprising . |
trees are ubiqutous structures which appear naturally in a large number of physical , chemical , biological and social phenomena , such as river networks , diffusion limited aggregation , pulmonary arteries , blood vessels and tree species , social organizations , decision structures , etc
. they also play an important role in computer science ( use of registers and computer languages ) , in graph theory , and in various methods of statistical physics such as cluster expansions and renormalization group . in spite of their apparent structural simplicity , and the large body of scientific work on trees
( a sample of which is found in @xcite-@xcite , @xcite,@xcite-@xcite and references therein ) , they still offer challenges even related to the quantitative description of their topological structure . at the dawn of the science of complex networks @xcite , it is therefore rather important to have a complete understanding of all the tree structures and their properties .
a tree is defined as a set of points ( vertices , nodes ) connected with line segments ( branches , or edges ) such that there are no cycles or loops ( a connected graph without cycles ) . for the simplest ( unlabeled ) rooted plane binary tree ,
each vertex has exactly three connecting branches , except for one vertex which is distinguished from all the others by having only two connecting branches coined as the root ( r ) of the tree , and a certain number of vertices with a single connecting branch called the ` leaves ' .
the height of the tree is defined by the maximum number of levels starting from the root ( which has height 0 ) , and it can be calculated as the maximum number of branches one has to pass to reach the root from its vertices ( since the leaves have only one branch , it means that this longest excursion must start from one of the leaves ) .
the paths from the leaves to the root define a natural direction on the tree ( similarly to the river flow ) which is always towards the levels of lower height . a tree of height @xmath2 we call _ complete _ , if it has @xmath3 leaves each being a distance @xmath2 from the root .
let us now mention three applications of the mathematics of trees which are directly connected to the so - called horton - strahler index of the tree , which is the subject of interest of the present paper .
originally , the horton - strahler index of a binary tree was introduced in the studies of natural river networks by horton @xcite and later refined by strahler @xcite , as a way of indexing real river topologies , since river networks are topologically similar to binary trees . by definition ,
a leaf has a rank of 0 ( some authors associate the value of 1 ) , and a vertex has a rank of @xmath4 where @xmath5 is the index function with @xmath6 and @xmath7 being the ranks of the two connecting vertices from the level above . when @xmath8 the index is called the horton - strahler index ( hs ) .
the quantity of particular interest is the hs index of the root which thus categorizes the topological complexity of the whole tree .
several other quantities can be introduced in relation to the hs index .
segment _ of order @xmath2 @xcite , or a _ stream _ of order @xmath2 @xcite is a _
maximal _ path of branches connecting vertices of hs index @xmath2 , ending in a vertex with index @xmath9 .
let @xmath10 denote the number of segments of order @xmath2 of a tree @xmath11 with @xmath1 leaves , and @xmath12 is the average physical length of a segment of order @xmath2 ( the average @xmath13 is taken on the tree @xmath11 ) .
the _ bifurcation ratios _ @xmath14 are defined as @xmath15 , and the length ratios via @xmath16 . horton and strahler have empirically observed that for river networks both the @xmath17 and @xmath18 tend to approximate a geometric series , @xmath19 with @xmath20 and @xmath21 with @xmath22 .
such networks are called _ topologically self - similar _ @xcite .
the notion of hs index is further refined by introducing the _ biorder _ @xmath23 of a vertex , representing the hs indices of its two children @xcite,@xcite , @xcite , and then studying the ramification matrix , with elements related to the number of vertices with a given biorder .
another interesting application of the mathematics of binary trees and the hs index , is in the description of the branched structure of diffusion - limited aggregates see ref . @xcite and references therein . in this case
the structures are grown on a substrate ( which can be a point or a plane ) by letting small particles diffuse towards the aggregate where they stick indefinitely at their point of first contact with the cluster .
this creates complex and involved branched structures , whose topological complexity still remains a challenging problem to describe .
finally , the last application we would like to mention is known as the _ word bracketing problem _ @xcite which has obvious implications in computer science .
let us consider an alphabet of @xmath1 letters , @xmath24 and a word @xmath25 , @xmath26 .
a 2-bracketing of the word @xmath27 is a partition of its letters ( by keeping their order ) in groups of two units enclosed in brackets , where a unit can be a letter or a subpartition enclosed in brackets , such as @xmath28 , or @xmath29 , etc .
the bracket @xmath30 between two units may be associated with a multiplicative composition law @xmath31 .
for example let the alphabet @xmath32 be all the positive integers , and the composition law be the regular multiplication of numbers .
then a bracketing of the multiple product @xmath27 corresponds to one particular way of calculating @xmath27 .
a one - to - one correspondence can be made immediately to trees : let the letters @xmath33 of the word @xmath27 be associated with the leaves of a binary tree . to a particular bracketing of @xmath27 it corresponds a particular tree constructed by associating a lower level vertex to a bracketing @xmath30 ( one may think of the brackets as representing the branches of the tree ) . the main question is how many ways are there to calculate such a product . if one assumes that the multiplication law is neither associative nor commutative , then the problem is refered to as the catalan problem , see ref .
@xcite for a number of solutions .
the number of such bracketings is given by the catalan numbers , @xmath34 .
the corresponding set of trees ( see fig.1 for @xmath35 ) is in fact the set of rooted , unlabeled binary plane trees according to this bijection .
= 5.0 in for later reference , we mention that the generating function @xmath36 of the catalan numbers obey the equation @xmath37 , with @xmath38 , so @xmath39 the power series @xmath40 converges within a disk of radius @xmath41 .
the problem of enumerating trees becomes more difficult if the composition law is commutative , which was first studied by wedderburn and etherington ( we ) @xcite , @xcite , @xcite . in the tree language , this means that two trees are considered identical if after a number of successive reflections with respect to the vertical axes passing through the vertices they can be transformed into each other and in this case they are said to be homeomorphic @xcite .
for the example shown in fig .
1 , there are only two such trees , since trees 1 ) , 2 ) , 4 ) and 5 ) can be transformed into each other .
the trees that can not be transformed into each other are called non - homeomorphic .
the set of non - homeomorphic trees is called the set of _ ambilateral _ trees , @xcite , @xcite .
let the number of such trees with @xmath1 leaves be denoted by @xmath42 .
the generating function ( gf ) defined as @xmath43 obeys the nonlinear functional equation : @xmath44 which has extensively been studied by wedderburn @xcite .
otter @xcite studying a more general counting problem where the vertices can have at most @xmath45 branches comes to the conclusion that for the ambilateral trees , if @xmath1 is large we have : @xmath46 where @xmath47 .
the method developed by otter gives an iterative approach to @xmath48 and @xmath49 .
for example @xmath49 is @xmath50 where @xmath51 , @xmath52 so that for @xmath35 one already obtains an extremely close value of @xmath53 .
later , bender developed a more general approach @xcite deriving the same results .
the coefficient @xmath48 in ( [ wen ] ) can also be computed : @xmath54 .
the more practical application of the bracketing problem within computer science is the computation of arithmetic expressions by a computer .
a general arithmetic expression involving only binary operators can simply be mapped onto a binary tree , called the syntax tree , which has as leaves the operands and the inner vertices the operators .
a computer traverses this tree from the leaves towards the root and it uses registers to store the intermediate results .
in general there are many ways of traversing such a tree , and the program that uses the _
minimal _ number of registers is the most efficient , or optimal one .
ershov has shown @xcite that the optimal code will use exactly as many registers to store the intermediate results as the hs index of the associated syntax tree . in the present paper
we investigate how the hs index is distributed on both the rooted , unlabeled , plane binary set of trees , and on the ambilateral set of binary trees .
we first answer this question on the rooted , unlabeled , plane binary set , since it is simpler , but it will also provide us with a technique that can be extended to tackle the problem for the ambilateral set .
for this set , the question was first answered by flajolet , raoult and vuillemin , @xcite with a method somewhat similar to the one presented here .
the enumeration problem of the hs index on the ambilateral set is , however , inherently more difficult since it involves functional equations with nonlinear dependence in the argument similar to eq .
( [ nlf ] ) , and therefore an explicit solution in a closed form becomes impossible to attain .
the derivation of an approximant formula for the number of ambilateral trees sharing the same hs index at the root is the main result of this paper .
the paper is organized as follows : first we present our derivation of the enumeration problem for the hs index on the unlabeled set in section ii , and then use this method of derivation from this case to develop a technique that can be used to attack the enumeration problem on the ambilateral set in the asymptotic limit , presented in section iii .
section iv is devoted to conclusions and outlook .
let us observe that the root @xmath55 of the tree has always two subtrees attached to it via the two branches , with @xmath2 and @xmath56 leaves , respectively , @xmath57 .
let @xmath58 denote the number of unlabeled trees with @xmath1 leaves that share the same hs index @xmath59 at the root .
a recursion is found for this number in the light of the observation above : @xmath60 with the conventions @xmath61 , @xmath62 , @xmath63 .
if the generating function for the variable @xmath1 is defined as @xmath64 , then it obeys : @xmath65 next we give an exact solution to ( [ recc ] ) .
let us introduce the sum @xmath66 then @xmath67 , and after rearranging the terms , eq .
( [ recc ] ) becomes @xmath68 , where @xmath69 .
this means that @xmath70 , i.e. , : @xmath71 note that the left hand side of ( [ b ] ) remains invariant to @xmath72 which is another solution of ( [ b ] ) .
however , since in case of the hs index @xmath73 , this latter solution has to be dropped .
if we make @xmath74 , ( [ b ] ) simplifies to @xmath75 . which , after dividing on both sides by @xmath76 , and introducing @xmath77 , becomes : @xmath78 let us now write @xmath79 , such that @xmath80
. then ( [ z ] ) becomes @xmath81 which leads to @xmath82 , @xmath83 , and which in turn is solved easily .
thus , @xmath84 , so one finally obtains : @xmath85 eq .
( [ ux ] ) is the exact solution to ( [ recc ] ) in the complex @xmath86 plane . on the real axis , within the radius of convergence @xmath87
the above expression takes the form : @xmath88 $ ] , @xmath89 . since within the convergence disk
one must have @xmath90 , we just obtained the identity ( using ( [ a ] ) ) : @xmath91 this identity can be checked to hold via more direct methods @xcite .
the singularities of @xmath92 lie on the positive real axis at : @xmath93 with an additional singularity at infinity ( corresponding to @xmath94 ) .
we certainly have @xmath95 . on the other hand
if one simply iterates ( [ recc ] ) we obtain : @xmath96 where @xmath97 is a polynomial in @xmath86 of order @xmath98 : @xmath99 , @xmath100 , @xmath101 , ... etc .
one can find an explicit form for this polynomial from the general solution ( [ ux ] ) if one invokes the identity @xcite : @xmath102 , so that ( [ pol ] ) is recovered with : @xmath103 .
it is easy to show , however , that @xmath104 , so the polynomial simplifies to : @xmath105 expression valid on the whole complex @xmath86 plane . based on the explicit solution we obtained , one can give an exact form to the distribution of the hs index on the unlabeled set of trees , by inverting the generating function via : @xmath106 one can write : @xmath107 where @xmath108 . by cauchy
s theorem the integrals in ( [ cauchy ] ) are readily performed , and one obtains : @xmath109^{-(n-2^r+1)},\ ; & n\geq 2^r \\ 0 , & \!\!\!\!\!0 \leq n
\leq 2^r-1 \end{array } \right.\;\;\;\;\label{final0}\end{aligned}\ ] ] from ( [ apart ] ) it follows that @xmath110 . to obtain the last equality we used the form ( [ prod ] ) and ( [ sing1 ] ) .
thus : @xmath111 the numbers @xmath112 can be calculated as follows : observe that @xmath113 where we used the lhpital rule in the last equality . on the other hand from ( [ pol ] ) and ( [ ux ] )
it follows : @xmath114 . taking the derivative of this equation at the point @xmath115 , and inserting it in ( [ aaj ] ) it yields : @xmath116^{2^r-1}}\ ] ] thus
, we obtain from ( [ final0 ] ) the more explicit form @xmath117^n } , \;\;\;n\geq 2^r , \label{explicit}\ ] ] or using ( [ sing1 ] ) : @xmath118^{2n-2 } \label{final1}\ ] ] an expression first derived by flajolet et .
al . @xcite . following this paper @xcite
, our polynomials @xmath119 can be simply connected to the tchebycheff polynomial @xmath120 @xcite , via the relation : @xmath121 . if one employs the poisson resummation formula for functions defined on a compact support ( see appendix b in ref .
@xcite ) on ( [ final1 ] ) , an equivalent combinatorial expression can be derived in the form : @xmath122(k)\big|_{k=1+(2m-1)2^r}$ ] , where @xmath123 is the finite difference operator . for a different method ,
see @xcite .
_ scaling limits .
_ next we briefly present the results of an asymptotic analysis on the @xmath58 numbers .
since @xmath58 is an enumeration result , it typically contains several scaling limits . in physical processes , during the growth of branched structures , usually only one of these limits is selected , and in frequent cases this limit has self similar properties ( such as for dla , or for random generation of binary trees , @xcite ) . by definition , the family of trees that obey @xmath124 } ) / r = const .
\equiv \ln{{\cal b}}$ ] is called topologically self similar @xcite , where @xmath125 is the _ bifurcation number_. + 1 ) @xmath126 and @xmath59 fixed . in this case the first term in ( [ final1 ] ) dominates the sum and the asymptotic behavior is given by @xmath127}}$ ] .
the rate of the exponential growth is a number between @xmath128 and @xmath129 .
2 ) @xmath130 , @xmath131 , @xmath132 . here
the first term in ( [ final1 ] ) is still dominant ( the rest being exponentially small corrections ) and yields : @xmath133 .
if @xmath134 diverges with @xmath59 slower than exponential , we have topological self similarity with @xmath135 .
+ 3 ) @xmath130 , @xmath131 , @xmath136 , with some @xmath137 . in this case
the rest of the terms in ( [ final1 ] ) ( after the first has been factored out ) are of the type @xmath138 and the final expression is : @xmath139 .
the topological self similarity is obvious with @xmath135 .
the factor @xmath140 is given by @xmath141 .
+ 4)@xmath130 , @xmath131 , @xmath142 , and @xmath143 . in this case
the analysis is performed easier from the combinatorial expression of @xmath58 and yields : @xmath144 .
let us now analyze the same question on the set of ambilateral trees , and denote the number of ambilateral trees with @xmath1 leaves and hs index @xmath59 by @xmath145 .
we certainly must have the relation @xmath146
= 3.0 in the table in fig .
[ tbl1 ] gives the distribution of the hs index for @xmath1 up to 32 and @xmath147 .
we can check easily that @xmath148 , and @xmath149 , so for simplicity these are not represented in the table . the numbers @xmath145 obey slightly more complicated recurrence relations since now the counting has to be done on a more restricted set .
we must distinguish between odd and even @xmath1 values .
however , the two cases can be combined into one , if the convention @xmath150 for @xmath151 non - integer is adopted .
the corresponding recurrence relation becomes : @xmath152 + m^{(r)}_{n/2 } \sum_{s=0}^{r-1 } m^{(s)}_{n/2 } + \frac{1}{2 } m^{(r-1)}_{n/2}\big ( 1+m^{(r-1)}_{n/2}\big)\end{aligned}\ ] ] the generating function @xmath153 will thus obey : @xmath154 ^ 2 + v_{r-1}(\xi^2 ) } { 1-\sum\limits_{s=0}^{r-1 } v_s(\xi ) } , \;\;r\geq 1 , \label{rec2}\ ] ] and @xmath155 . as a check for the correctness of ( [ rec2 ] ) ,
let us see if we recover the identity @xmath156 ( which follows from ( [ closure ] ) ) .
( [ rec2 ] ) is equivalent to @xmath157 ^ 2 + v_{r-1}(\xi^2 ) $ ] .
introduce the temporary variable @xmath158 and sum both sides of the equation over @xmath59 , @xmath159 .
one obtains @xmath160 ^ 2 + g(\xi^2)$ ] . using the identity @xmath161 , one finds @xmath162 ^ 2 + \frac{1}{2 } g(\xi^2)$ ] which is precisely eq .
( [ nlf ] ) , showing that @xmath163 , i.e. , the relation @xmath164 holds , indeed .
in contrast to the previous case , the functional recurrence ( [ rec2 ] ) can not be treated in an exact analytical fashion due to the functional dependence on @xmath165 . however , one can derive the asymptotic behaviour and make statements that will lead to rather close approximations of the @xmath145 numbers .
it is instructive to look at a few particular values , first : @xmath166 inverting @xmath167 , one obtains : @xmath168/6 $ ] , @xmath169 , which can be checked to hold , see the table in fig .
the result from the inversion of @xmath170 is already so complicated that it is not worth presenting .
as the index @xmath59 increases , the polynomial expressions become more and more involved .
figure [ fig4 ] shows the function @xmath171 in the interval @xmath172 $ ] .
= 4.0 in for every @xmath59 , the power series for @xmath173 has non - negative coefficients , @xmath174 .
based on a classic theorem of complex analysis , this means that on the circle of convergence , of radius @xmath175 , there will be a singularity of @xmath173 at @xmath176 .
next we show , that we have the ordering @xmath177 for @xmath178 , and the limit @xmath179 exists and it is equal to @xmath180 .
we shall use mathematical induction to prove the ordering . from the particular examples above it follows that @xmath181 , @xmath182 .
let us now assume that @xmath183 for all @xmath184 , @xmath185 .
we will show that @xmath186 .
note that the radius of convergence for @xmath187 is @xmath188 , if @xmath189 is less than unity . by _
reductio ad absurdum _
, let us assume first , that @xmath190 .
this means , that @xmath191 is analytic in @xmath192 . from ( [ rec2 ] ) , @xmath193 by the argument above
, @xmath194 is analytic in @xmath192 ( its radius of convergence is @xmath195 , since by assumption @xmath196 ) . in the denominator of ( [ 4rp1 ] ) , _ all _ functions @xmath197 ,
@xmath198 are analytic in @xmath192 , because by assumption they all have radii of convergence strictly larger than @xmath192 .
however , @xmath199 is singular in @xmath192 , and the singularities do not cancel in the numerator and denominator of ( [ 4rp1 ] ) , and thus @xmath200 is singular in @xmath192 , a contradiction .
we are left to prove that @xmath201 can not hold .
let us denote @xmath202 .
again , we assume , that @xmath201 is true .
it is easy to show , that for any _ finite _ @xmath59 , @xmath203 .
this means from the recurrence relation that @xmath204 ( in the numerator of ( [ rec2 ] ) we have only functions analytic at @xmath192 ) . since @xmath205/[1-b_r(\xi)]$ ]
, from the assumption @xmath201 it would follow that the equation @xmath206 _ can not _ have any solutions ( @xmath200 is analytic within the circle of convergence ) in the interval @xmath207 .
( note that in the interval @xmath207 , the numerator @xmath208 can not be zero , since the power series @xmath199 has only positive coefficients ) .
the equation @xmath206 is equivalent to @xmath209 .
however , from ( [ rec2 ] ) @xmath210/v_r(x)$ ] , thus , the equation @xmath211 should have no solution in @xmath207 . if @xmath212 is arbitrarily close to @xmath192 , then @xmath213 is arbitrarily large .
however , since @xmath214 , @xmath215 , and @xmath216 are both bounded from above .
thus , for @xmath212 sufficiently close to @xmath192 , we have @xmath217 . on the other hand
, the hs index of a tree @xmath11 equals to the height of the largest , complete , balanced tree embedded in @xmath11 .
this means , that @xmath218 for @xmath219 .
also , @xmath220 .
in other words , one must have @xmath221 .
= 4.0 in this means that @xmath222 , @xmath223 , and @xmath224 . since @xmath225
, there will always be an @xmath226 , ( @xmath227 ) , sufficiently close to zero , such that @xmath228 .
therefore , there must exist an @xmath207 , for which ( [ equ ] ) holds , which is a contradiction .
thus , we have proven that @xmath177 , for all @xmath178 . as a matter of fact
we have also shown , that the radii of convergence satify : @xmath229 since the series @xmath192 is monotonically decreasing , and bounded from below , the limit @xmath230 exists .
we have shown that @xmath231 .
since the radius of convergence for the left hand side is the minimum of all the radii of the terms in the summation , i.e. , @xmath232 , it must equal to the radius of convergence for @xmath233 , which , as shown by otter and bender is @xmath234 , @xmath235 . taking the limit @xmath131 in ( [ aneq ] ) , we get @xmath236 ( since by definition @xmath237 , so @xmath238 ) . or , using ( [ nlf ] ) :
@xmath239 an identity also shown by bender .
( [ iter ] ) and ( [ nlf ] ) can simply be combined to give the iterative computation of @xmath232 in the form already mentioned in the introduction , as follows : if we make the temporary notation @xmath240/\sqrt{\xi}\ ; , \label{uu}\ ] ] eq .
( [ nlf ] ) takes the form @xmath241 and eq .
( [ iter ] ) simplifies to @xmath242 let @xmath243 .
then , from ( [ si1 ] ) @xmath244 , or @xmath245 , where there are a total of @xmath1 compositions for the @xmath27 function , @xmath1 arbitrary .
let us now choose @xmath246 .
this means , @xmath247 , by virtue of ( [ si2 ] ) . from ( [ uu ] ) , @xmath248/\alpha^{2^{n}}$
we have shown previously , that @xmath249 ( it is the limit of the monotonically decreasing series @xmath250 ) , therefore we have : @xmath251 since @xmath252 , and where @xmath253 , @xmath254 , just as in the introduction . the convergence is double - exponential , very fast .
as in section ii , the asymptotic behavior of the @xmath145 numbers for relatively large @xmath1 and @xmath59 is governed by the innermost singularity of @xmath173 on the real axis .
the graph of @xmath255 shown in figure [ fig4 ] suggests , that the generating function is in fact well behaved in a certain interval to the right of the radius of convergence , @xmath256 , see also figure [ fig5 ] .
the existence of this interval comes from the fact that the singularities of the term with nonlinear argument @xmath257 in the numerator of ( [ rec2 ] ) kick in only beyond the circle of convergence of @xmath257 , which is @xmath258 .
thus , in the interval @xmath259 the term with the nonlinear argument is analytic , which ultimately is responsible for this nice behaviour . because , @xmath260 , for convenience we shall define the interval of this nice behaviour to be @xmath261 .
in order to exploit this observation , we shall first rewrite the recurrence relation ( [ rec2 ] ) .
let us denote @xmath262 . with this notation , ( [ rec2 ] )
takes the form @xmath263 , @xmath264 , where @xmath265 .
this leads to the new recurrence : @xmath266 @xmath267 .
this would be exactly solvable if it were not for the dependence on the nonlinear argument @xmath165 .
note the resemblance to ( [ b ] ) .
let @xmath268 , which is an analytic function in @xmath269 .
we also have @xmath270 , the latter equality being shown previously .
this shows , that in the interval @xmath269 , the @xmath59-dependence _ weakens extremely fast _ , double - exponentially with increasing @xmath59 . as a matter of fact
, an upper estimate is @xmath271 in particular , @xmath272 , @xmath273 , @xmath274 , @xmath275 , @xmath276 , etc . therefore , from the point of view of the asymptotic behavior ,
the @xmath277 functions can be replaced by their asymptotic expression ( as @xmath131 ) : @xmath278 figure [ fig6 ] shows the functions @xmath277 on the interval @xmath269 for @xmath279 .
= 4.0 in thus , instead of eq .
( [ rec3 ] ) we will consider : @xmath280 the recurrence ( [ app ] ) in turn is easily solved in the way shown in section i. the result is : @xmath281 where @xmath282 for the moment is an arbitrary ( positive integer ) index .
recurrence ( [ app ] ) will become a good approximation to the recurrence ( [ rec3 ] ) from an index @xmath282 on .
the larger @xmath282 is the more accurate the approximation .
recurrence ( [ app ] ) is applied then with initial condition @xmath283 , which for modest @xmath282 values can be obtained by iterating ( [ rec3 ] ) @xmath282 times . what is the error we make when one replaces @xmath284 with @xmath285 on @xmath269 ? summing the differences ( [ diffu ] ) from @xmath286 to infinity , one obtains the estimate : @xmath287 .
thus , for example , @xmath288 is smaller than @xmath289 , @xmath290 is smaller than @xmath291 , etc .
therefore , we can finally write on @xmath269 : @xmath292 in fig .
( [ figxy ] ) we plot the rhs of ( [ apo ] ) and the @xmath199 function from iterating ( [ rec2 ] ) .
note that the approximation is very good , and it becomes virtually indistinguishable from the true function the closer @xmath86 is to @xmath232 .
larger @xmath282 values will also give better approximations , since the approximation is only applied from the @xmath282 index on .
however , @xmath282 can not be taken too high for approximation purposes , since it assumes that the exact expression of @xmath293 ( or @xmath294 ) is known .
this makes only the modest @xmath282 values ( less than 5 ) useful .
on the other hand , expression ( [ apo ] ) is very practical in analysing the singularities of @xmath295 and give rather close approximant expressions to these singularities .
in particular , we see that within the interval @xmath269 , ( [ apo ] ) preserves the property that if @xmath296 is a singularity of @xmath297 ( or a zero of @xmath298 ) then it is a singularity of @xmath299 ( or a zero of @xmath300 ) , whenever @xmath301 .
if one is interested in the asymptotic behavior , then a more tractable expression can be derived for the rhs of ( [ apo ] ) : the function @xmath285 is analytic on the interval @xmath269 , and since already for modest @xmath59 values , the innermost singularity of @xmath199 ( denoted @xmath192 ) is extremely close to @xmath232 , one can safely replace @xmath285 in this neighborhood by : @xmath302 .
= 4.0 in this leads to the approximant : @xmath303 for sufficiently large @xmath59 ( here `` large '' means @xmath304 ) where @xmath305 next , we compute @xmath306
. one can use a very similar method to the one employed to obtain ( [ afa ] ) , to give : @xmath307 so , @xmath308 . if one computes @xmath309 for @xmath310 , we have @xmath311 , and thus @xmath312 .
if we were to use @xmath313 , then one would obtain @xmath314 , so @xmath315 and slightly improve the approximation on @xmath309 .
no significant improvement will be obtained with larger @xmath282 values .
figure [ fig9 ] shows the agreement of the form given in ( [ apo1 ] ) . for clarity , we defined the function @xmath316 given by : @xmath317 here we use the true @xmath199 function using numerical iteration of ( [ rec2 ] ) , and evaluate it in the points @xmath318 .
if the approximation ( [ apo1 ] ) is good , then one should have @xmath319 . as seen from fig .
[ fig9 ] the approximation is already excellent for @xmath320 close to @xmath232 ( which corresponds to the @xmath321 point in this plot ) .
the interval @xmath269 in these transformed corrdinates corresponds to @xmath322 .
there are no fitting parameters , we used for @xmath323 and @xmath309 the values derived above . in order to obtain the approximation to the _ number _ @xmath145 of ambilateral trees with the same hs index at the root , we will have to invert ( [ apo1 ] ) .
the singularities of the rhs of ( [ apo1 ] ) are given by : @xmath324 ( at the moment we do not care whether some of the singularities will fall outside the interval @xmath269 , we just simply want to invert ( [ apo1 ] ) , and then at the end keep only those terms from the final expression that were generated by the singularities within @xmath269 ) .
= 4.0 in in a similar manner to the previous section , we first bring @xmath325 to an inverted polynomial form : @xmath326^{2^r } } { 2^{r+1 } \theta^{2^{r+1}-1 } q_r(\xi)}\ ] ] where @xmath327 is the polynomial : @xmath328 .
the case from the previous section ii corresponds to @xmath329 and @xmath330 .
thus , if we denote by @xmath331 the numbers coming from the inversion of @xmath332 , then : @xmath333^{2^r } } { 2^{r+1 } q_r(\xi)}\ ] ] we have : @xmath334 after performing the integrals , one obtains : @xmath335^{-n-1 } \sum_{m=0}^{\min\{n,2^r\ } } { 2^r \choose m } ( 1-\alpha \theta^2)^{2^r - m } \left[\theta^2 \xi_j^{(r ) } \right]^{m } \label{ems}\ ] ] this expression shows that the @xmath331 may only _ approximate _ the @xmath336 numbers in a certain limit .
this is seen from the fact that while one must have @xmath337 for @xmath338 , and @xmath339 , this is not respected by ( [ ems ] ) ( it would only be respected if @xmath340 , however , this is not the case , and the reason behind this discrepancy is the neglected nonlinearity from the calculations ) .
the limit , in which the approximation becomes good is for @xmath59 large ( it means @xmath341 ) and @xmath342 . in this case
the sum over @xmath45 can be performed , and one obtains : @xmath335^{-n-1 } \left [ 1+\theta^2 ( \xi_j^{(r)}-\alpha ) \right]^{2^r } \label{ems1}\ ] ] the @xmath343 numbers can be calculated in exactly the same way we did in the previous section .
this leads to : @xmath344^{2^r-1}}\;.\ ] ] inserting it into ( [ ems1 ] ) it yields : @xmath345 \left(\xi_j^{(r ) } -\alpha \right ) } { [ \xi_j^{(r)}]^{n+1 } } \label{mbar}\ ] ] as a check to the correctness of ( [ mbar ] ) we can take @xmath346 and @xmath347 from the unlabeled case , to obtain ( [ explicit ] ) . equation ( [ mbar ] ) explicitely shows the contribution of each singularity .
however , if we want to approximate the @xmath145 numbers , we should also account for the condition @xmath348 . using the expression ( [ sing2 ] ) , this leads to @xmath349 , where : @xmath350 thus , using again ( [ sing2 ] ) : @xmath351 } ( -1)^{j+1 } \frac{\tan^2\!{\left(\frac{j\pi}{2^{r+1}}\right ) } \left [ 1 + \tan^2\!{\left(\frac{j\pi}{2^{r+1}}\right)}\right ] } { \left [ \alpha + \theta^{-2 } \tan^2\!{\left(\frac{j\pi}{2^{r+1}}\right)}\right]^{n+1 } } \label{mbart}\ ] ] when the asymptotic limit is generated by the innermost root @xmath352 , i.e. , by the first term in ( [ mbart ] ) , one obtains for the topologically self similar ambilateral trees , the scaling behaviour : @xmath353 and therefore @xmath354 .
let us now see how well formula ( [ mbart ] ) approximates the @xmath355 numbers . to do this , we shall define the error @xmath356 \cdot 100\% $ ] .
for example from the table in fig .
[ tbl1 ] , @xmath357 .
the formula above gives @xmath358= 445781858 $ ] , and thus @xmath359 .
further error values : @xmath360 , @xmath361 , @xmath362 .
combinatorial enumeration of trees is typically difficult to solve when the set under enumeration obeys symmetry - exclusion principles , such as for the ambilateral case treated here .
these symmetry - based constraints may arrise in realistic situations and thus forces us to enumerate _
classes _ of subsets of trees . in the ambilateral case a class is defined as being formed by those binary trees that have the same number of leaves and hs index at the root and can be obtained one from another via successive reflections with respect to the nodes of the tree .
certainly , the symmetry operation defining the class must be an invariant transformation of the topological index ( hs in our case ) .
an other example of such symmetry - operation - generated class - enumeration is the case of the `` leftist trees '' playing an important role in the representation of _ priority queues _ , first shown by crane @xcite , followed by knuth @xcite , who gives their explicit definition .
an elegant enumeration for the leftist trees , using generating function formalism was only given very recently by nogueira @xcite .
the existing solutions to such class - enumerations on trees ( such as ours and that of flajolet et .
al . @xcite and of nogueira @xcite ) are obtained via methods taylored for the particularities of the set and symmetry operation in question .
it is desirable to have , however , at least on a formal level , a general encompassing theory of class - enumerations of topological indices . in this direction ,
powerful methods such as that of the antilexicographic order method developed by erds and szkely @xcite , or the method of bijection to schrder trees developed by chen @xcite may turn to be effective after a suitable extension to include topological indices such as the horthon - strahler index .
this , however , stands as an open problem .
i am especially thankful to eli ben - naim for introducing this problem to me , and for the many constructive suggestions while i was working on it .
useful discussions and comments from i. benczik , t. brown , w. y. c. chen , p. l. erds , m. hastings , g. istrate and r. mainieri are also gratefully acknowledged .
this work was supported by the department of energy under contract w-7405-eng-36 . | the horton - strahler ( hs ) index @xmath0 has been shown to be relevant to a number of physical ( such at diffusion limited aggregation ) geological ( river networks ) , biological ( pulmonary arteries , blood vessels , various species of trees ) and computational ( use of registers ) applications . here
we revisit the enumeration problem of the hs index on the rooted , unlabeled , plane binary set of trees , and enumerate the same index on the ambilateral set of rooted , plane binary set of trees of @xmath1 leaves .
the ambilateral set is a set of trees whose elements can not be obtained from each other via an arbitrary number of reflections with respect to vertical axes passing through any of the nodes on the tree . for the unlabeled set we give an alternate derivation to the existing exact solution . extending this technique for the ambilateral set , which is described by an infinite series of non - linear functional equations , we are able to give a double - exponentially converging approximant to the generating functions in a neighborhood of their convergence circle , and derive an explicit asymptotic form for the number of such trees . epsf |
if supersymmetry ( susy ) exists at the electroweak scale , then gluinos and squarks will be copiously produced in pairs at the lhc and will decay via cascades involving other susy particles to the lightest susy particle ( lsp ) , the @xmath4 . in most models
the @xmath4 is stable , must be neutral and therefore escapes the detector .
it should then be easy to observe deviations from the standard model such as an excess of events with multiple jets plus missing energy @xmath5 or with like - sign dileptons @xmath6 plus @xmath5@xcite .
determining susy masses is more difficult because each susy event contains two lsp s , and there are not enough kinematic constraints to determine the momenta of these .
the strategy developed in ref.@xcite and in this paper involves three steps .
first , we use a simple inclusive analysis to establish a deviation from the standard model .
we select events with at least four jets and large missing energy and plot the distribution of @xmath7 typically , this is dominated by standard model processes at low @xmath8 but is a factor of 510 larger than the standard model prediction for large @xmath8 . the value of @xmath8 at which the signal exceeds that standard model backgrounds provides a first estimate of the susy masses .
the second step in the strategy is to identify characteristic signatures of particles occurring near the end of the susy decay cascades and to use these as the starting point for further analysis .
this is best explained by an example .
suppose that gluinos are slightly heavier than squarks and that @xmath9 is kinematically allowed
. then one can have the following decay chain : @xmath10 such an event typically contains two hard jets from the @xmath11 decays , two @xmath12 jets from the @xmath13 decay , large @xmath5 , and soft jets from the gluino decays and from gluon radiation . in this case , one can reconstruct @xmath14 as a peak in the @xmath15 mass distribution and measure its mass@xcite . then the @xmath13 can be combined with either of the two hard jets .
the smaller of these two masses must be less than the squark mass and so has a sharp edge that measures a known function of the @xmath16 , @xmath17 , and @xmath18 masses .
in many cases several such measurements can be made to determine several combinations of masses more or less precisely .
given actual data , the third step would be to make a global fit of a susy model to all available measurements , including both the precision measurements just described and more inclusive ones such as the jet , lepton , and @xmath12-jet multiplicities and @xmath19 distributions .
such an analysis involves simulating large numbers of signal samples and is beyond the scope of this study .
instead , we try to determine the susy parameters using just the precision measurements of combinations of masses . in some cases
this almost completely determines the susy model , while in others it significantly constrains it .
.sugra parameters for the five lhc points.[params ] [ cols="^,^,^,^,^,^ " , ] if @xmath20 is near its kinematic limit , then the velocity difference between the @xmath21 pair and the @xmath4 is minimized in the rest frame of @xmath22 .
having both leptons hard requires @xmath23 . assuming this and @xmath24 implies that the endpoint in figure [ c5_mll ] is equal to the @xmath4 mass .
an improved estimate could be made by detailed fitting of all the kinematic distributions .
events were selected with @xmath25 , and the @xmath4 momentum was calculated using this crude @xmath4 mass and @xmath26 the invariant mass @xmath27 of the @xmath21 , the highest @xmath19 jet , and the @xmath4 was then calculated and is shown in figure [ c5_mlljh ] .
a broad peak is seen near the light squark masses , 660@xmath28 .
this peak contains complementary information to that obtained from figure [ c5_mbbj ] and more information about the masses could be obtained by performing a combined fit .
a detailed analysis of the dilepton mass spectrum would require varying @xmath29 , @xmath30 , @xmath31 , and the @xmath32 distribution and fitting the distributions of @xmath33 , @xmath34 , @xmath35 , and @xmath36 . rather than do this ,
a sample of 50k events with @xmath37 but otherwise the same parameters was generated .
a comparison of some of the relevant masses is shown in table [ masses ] ; the slepton mass changes by @xmath38 , and the rest are essentially identical .
the @xmath33 mass distribution near the edge is shown in figure [ c5p20_mll ] .
there is a shift in the location of the edge by about @xmath39 , which should be observable .
there is not much change in the @xmath19 distribution of the @xmath40 pair , figure [ c5p20_ptll ] .
the most sensitive distribution is @xmath36 , where by definition @xmath41 .
there is a clear change in the shape , as one would expect . for fixed values of the other parameters , one ought to be able to determine @xmath42 to @xmath43 ,
although careful subtraction of standard model backgrounds will be necessary .
gluino decays at lhc point 5 have a sizable branching ratio to @xmath44 .
we have attempted to isolate such a sample by searching for top decays . in order to reduce the background from standard model top production ,
the cut on @xmath8 has been raised to @xmath45 for this analysis .
the event selection is as follows : * @xmath46 ; * @xmath47 jets with @xmath48 , @xmath49 ; * transverse sphericity @xmath50 ; * @xmath51 ; * @xmath52 .
* exactly two b - jets with @xmath53 .
if the invariant mass of the two-@xmath12 system is within @xmath54 gev of the higgs mass , the event is rejected . to search for @xmath55
, we calculate the invariant mass of all dijet combinations with both jets having @xmath56 gev and with neither jet being tagged as a b. combinations with @xmath57 mass are considered w candidates .
combinations where @xmath58 are used to model the shape of the background ; we will refer to these events as the @xmath59 sideband region .
the @xmath60 distribution is shown in figure [ c5-w - for - top ] where a clear @xmath59 peak is visible .
the average jet multiplicity in these events is @xmath61 and hence there is a large combinatorial background in this figure .
dijets in the @xmath59 mass and @xmath59 sideband regions are then combined with each of the two @xmath12 jets , giving two combinations for each @xmath59 .
results of this reconstruction for the @xmath59 signal and sideband regions are shown in figure [ c5-marjietopu ] .
subtraction of the sideband distribution from the signal yields figure [ c5-marjietop ] . a clear top signal is seen .
there is a small background of standard model production of top quarks , shown as the hatched histogram on this figure .
the dominant decay chain giving rise to this signal is @xmath62 the kinematics of the decay @xmath63 , @xmath64 restricts the @xmath65 invariant mass distribution to be between 203 and 356 gev .
there is some evidence in figure [ c5-marjietop ] for an excess of events in this region .
unfortunately , other possible combinations and decays such as @xmath66 , @xmath67 , @xmath68 , which has a smaller combined branching fraction , make the exact interpretation difficult .
thus , when discussing the determination of susy parameters in section [ sec : scan-5 ] , we assume that top decays in susy events can be seen at point 5 ( indicating that the @xmath69 channel is open ) , but we will not assume that the @xmath70 mass can be determined .
lhc points 1 and 2 have gluino masses of about 1 tev and squark masses about 50 gev lighter .
these are close to the upper limit of the expected range if susy is to be relevant to electroweak symmetry breaking .
the cross sections are quite small , so more than @xmath73 is needed for precision studies . only a few results will be presented here .
this analysis is very similar to the one for lhc point 5 , but since the signal cross sections are smaller , harder cuts are needed : * @xmath46 ; * @xmath47 jets with @xmath75 , @xmath76 @xmath77 ; * transverse sphericity @xmath50 ; * @xmath51 ; * @xmath52 .
events with exactly two tagged @xmath12 jets were then selected .
figures [ c1_mbb ] and [ c2_mbb ] show the @xmath15 mass distributions for points 1 and 2 respectively .
the light higgs would be discovered in this mode at either point with @xmath73 or less , although @xmath78 would still be needed to provide the most precise mass determination .
events passing these cuts and having a @xmath79 mass within @xmath80 of the peak were then selected .
these events were also required to have two and only two additional jets with @xmath81 .
this cut is quite inefficient , but attempts to improve its efficiency produced more background .
each of the two jets was combined with the @xmath79 pair .
the smaller of the two masses was selected for each event and plotted in figures.[c1_mbbj ] and [ c2_mbbj ] .
the endpoints for the decay sequence @xmath82 are 739 gev and 751 gev respectively .
measurement of this edge will be limited by statistics to @xmath83 for @xmath73 .
sleptons are quite heavy for lhc points 1 and 2 , so there is no edge in the dilepton mass spectrum as for point 5
. there is , however , a @xmath85 signal that can be used to distinguish these otherwise rather similar points .
the basic selection cuts given at the beginning of the previous subsection were applied .
in addition , the events were required to have two opposite - sign , same - flavor leptons with @xmath86 , @xmath87 , and @xmath88 in a cone @xmath89 .
the mass of the two highest @xmath19 such leptons is shown in figure [ c1_mll ] and figure [ c2_mll ] for lhc points 1 and 2 respectively .
the @xmath90 peak is questionable for point 1 but rather clear for point 2 , which has larger @xmath91 and hence more mixing of gauginos and higgsinos . even for point 2
, however , the statistical error on the number of @xmath90 s is @xmath92 .
clearly this is a measurement that needs higher luminosity .
there are also observable signals in the like - sign and in the opposite - sign , opposite - flavor dilepton channels
. these would certainly be useful in a global fit , but they do not seem to provide the sort of precise measurements being considered in this paper .
once a number of quantities have been measured , we can attempt to determine the particular susy model and the values of the parameters
. the strategy will be to attempt to perform a global fit to the model parameters using all of the available data , much as the standard model is tested using the @xmath59 and @xmath90 masses and the many quantities precisely measured by lep / slc .
such a fit is beyond the scope of our work , and we adopt a simpler procedure .
we assume that from measurements of global parameters such as those discussed in section [ sec : meff ] we know the approximate scale of the superpartner masses and have some idea that we might be in a sugra model .
the object is then to determine the parameters of that model and check its consistency .
we must therefore determine the parameters @xmath42 , @xmath93 , @xmath94 , @xmath91 and @xmath95 . as we will see @xmath94 is difficult to determine .
its value is given at the unification scale , and the value that is relevant for the phenomenology is the one evolved down to the electroweak scale .
many choices of @xmath94 evolve to the same ( fixed point ) value , so there is reduced sensitivity to its value at the unification scale .
in addition it always appears scaled by the yukawa coupling for the relevant quark or lepton .
hence its effect on lepton masses and on the quarks of the first two generations is very small .
its effect is significant for top squarks and at large @xmath91 for bottom squarks .
our strategy for determining the parameters is as follows .
we choose a point randomly in parameter space and compute the spectrum .
we assign a probability to this point determined from how well it agrees with our `` measured quantities '' using our estimates of the errors on those quantities .
the process is repeated for many points and the probabilities used to determine the central values of the parameters , their errors and their correlations .
the probability distribution functions are not always gaussian ; the @xmath96 ( @xmath97 ) errors quoted below are such that 15.87% of this distribution is below ( above ) the quoted value .
thus , 68.27% of the probability falls within our definition of @xmath98 . at lhc point 3 ( @xmath99 gev , @xmath100 gev ) , lep will discover @xmath13 and measure its mass .
the relevant error is that from theoretical calculations of the mass in the supergravity model which is likely to dominate the error from the lep measurement .
we will assume an error of @xmath101 .
using the results presented in section [ sec : lhc3 ] , we assume the following measurements and errors : * @xmath102 gev , * @xmath103 gev , * @xmath104 gev . as described above the mass difference @xmath105 is insensitive to the mass assumed for @xmath18 . using the strategy outlined above we get the following constraint on the parameters : * @xmath106 gev , * @xmath107 gev , * @xmath108 , * the sign of @xmath109 is determined to be @xmath110 , * @xmath94 is constrained to be greater than @xmath111 gev .
there are no clear correlations between the parameters .
the additional constraint that the average value of the light squark mass is within @xmath54 gev provides no additional restriction on the parameters .
the relevant phenomenological parameter is the value of @xmath112 evolved down to the the electroweak scale .
the relevant values are those for the third generation , @xmath113 and @xmath114 .
information on these can only come directly from data on bottom and top squarks . in this case
@xmath113 is not well constrained ; it is allowed to range from @xmath116 to @xmath117 gev . over this
range the mass of the two stop eigenstates varies only slightly from 270 and 320 gev at one end to 260 and 330 gev at the other .
constraining @xmath113 is very challenging as it appears scaled by the small factor of the bottom quark yukawa coupling
. the degree of precision may be surprising .
over most of the sugra parameter space @xmath17 and @xmath4 are gauginos ( _ i.e. _ , they have no higgsino components ) , @xmath118 then determines @xmath93 . @xmath119 and @xmath120 are then predicted and a consistency check of the model made by the measurement of @xmath121 . information on @xmath42 , is then obtained from @xmath122 and the value of @xmath123 is sufficient to constrain @xmath91 . the other measurements available at this point are now used to provide powerful consistency checks of the model .
the measurement of the bottom squark and other squark masses and the event rates for isolated leptons without jets discussed in section [ sec : lhc3 ] are examples .
another example is the branching ratio for @xmath124 .
figure [ branchfig ] shows this branching fraction for a selection of sugra models that have parameters in the slightly larger range @xmath125 gev , @xmath126 gev , @xmath127 . at point 3
this branching ratio can be constrained using the method described in section [ sub : br];. at lhc point 5 ( @xmath129 , @xmath130 , @xmath131 ) , @xmath13 will be discovered at lhc in its decay to @xmath132 from its production in the decays of supersymmetric particles , and its mass will be measured precisely from its decay to @xmath133 . using the results presented in section [ sec : lhc5 ] , we assume the following set of measurements : * @xmath134 gev , * the decay @xmath135 is allowed , * the end point of the spectrum in figure [ c5_mbbj ] is @xmath136 gev , * @xmath137 gev .
these results correspond to two possible solutions : * @xmath138 gev , * @xmath139 gev , * @xmath140 , * @xmath141 ; and * @xmath142 gev , * @xmath143 gev , * @xmath144 , * @xmath145 .
both of these solutions provide good fits to the `` data '' . while there is no constraint on @xmath112 ,
the values of @xmath114 and @xmath113 are constrained : @xmath146 ( @xmath147 ) gev and @xmath148 ( @xmath149 ) gev for @xmath109 positive ( negative ) .
the main differences between the mass spectra for these two solutions are in the masses of @xmath150 , @xmath151 , @xmath152 , and @xmath153 and the masses of the heavier higgs bosons ; they are significantly larger in the positive @xmath109 case .
the claimed sensitivity to the lepton decay spectrum ( see figure [ c5p20_pt2pt1 ] ) of @xmath154 gev implies that this parameter s range can be narrowed somewhat .
we have not investigated the sensitivity of this spectrum to the sign of @xmath109 . at this point
the errors used ( except for the one on @xmath123 ) are limited by statistics , so additional luminosity will cause the errors to drop . using 0.6 gev for the error on the dilepton endpoint and 23 gev for the error on the endpoint of figure [ c5_mbbj ] reduces the errors on @xmath93 and @xmath42 to @xmath155 and @xmath156 ( @xmath157 and @xmath158 ) respectively for positive ( negative ) @xmath109 ; the error on @xmath91 is not reduced . as well as the differences in the masses of the heavier higgs and gauginos noted above
, the negative sign solution has larger stop masses and hence a smaller branching ratio for @xmath159 .
we have investigated the sensitivity of the top quark signal discussed in section [ subsec : top ] to the sign of @xmath109 .
the parameters @xmath160 , @xmath161 , @xmath162 , @xmath145 , and @xmath163 gev have the opposite sign of @xmath109 from point 5 but give an excellent fit to the `` data '' used for fitting at this point . a sample of 100k events was generated for this parameter set , and the analysis leading to figure [ c5-marjietop ] was repeated .
the result is shown in figure [ c51-marjietop ] ; note that the bins in this figure are twice as wide as those in figure [ c5-marjietop ] .
it can be clearly seen that the amount of reconstructed top in susy events is reduced relative to that at point 5 and that this fact can be used to eliminate this alternative solution .
in addition , the solutions with @xmath145 have smaller branching ratios of @xmath164 so the observed number of higgs events should be able to severely constrain this case . at lhc point 4 ( @xmath165 , @xmath166 , @xmath167 ) , determination of the parameters
can not be done by the simple method already described . here
only two masses can be measured in a straightforward manner ; the light higgs mass from its decay to @xmath133 and the @xmath168 mass difference from the endpoint in the dilepton mass distribution .
we use these two measurements and the determination of the susy scale from the @xmath8 analysis of section [ sec : meff ] : * @xmath169 , * @xmath170 gev , * @xmath171 gev from the @xmath8 analysis .
these constraints restrict the parameter space to two regions : * @xmath172 gev , * @xmath173 gev , * @xmath174 , * @xmath112 is not constrained , * @xmath141 , * @xmath175 gev and @xmath176 gev ; or * @xmath177 , * @xmath178 , * @xmath179 gev , * @xmath112 is not constrained , * @xmath145 * @xmath180 gev and @xmath181gev . the uncertainty on @xmath182 is limited by statistics at low luminosity .
a reduction in its error would reduce the error on @xmath93 . in order to constrain parameters further , models whose parameters are consistent with these values
would need to be generated and their predictions for the distributions shown in figures [ c4_mllm0 ] , [ c4_mllemu ] , and [ c4_mllemudif ] calculated .
those which are inconsistent with the `` observed '' distribution can then be rejected .
this exercise is beyond the scope of this paper . at lhc point 1 and 2 ( @xmath183 , @xmath184 , @xmath72 ) , event rates are low and precision measurements difficult at low luminosity .
we use the following constraints .
* @xmath185 gev for point 1 , or * @xmath186 gev for point 2 , * @xmath187 gev from the @xmath8 , * the end point of the spectrum in figure [ c5_mbbj ] is @xmath188 gev . in the case of point 1
, there are two solutions : * @xmath42 unconstrained , * @xmath189 gev , * @xmath190 , * @xmath141 , * @xmath112 is not constrained , * @xmath191 gev and @xmath192gev ; and one with negative @xmath109 that has instead * @xmath193 gev , * @xmath194 . in the case of point 2 , there is again a solution for either sign of @xmath109 : * @xmath42 unconstrained , * @xmath195 gev , * @xmath196 , * @xmath197 , * @xmath112 is not constrained , * @xmath198 gev and @xmath199gev .
the lack of a constraint on @xmath42 at these points is alarming but can be explained .
@xmath93 is large and the renormalization group scaling from the gut scale forces the squark masses to be comparable to the gluino mass almost independent of the input value of @xmath42 . in cases of this type one needs to measure slepton masses which are less affected since their renormalization group scaling is controlled by @xmath200 rather than @xmath201 . since
the slepton masses are of order 500 gev , this is a difficult task .
even a lower bound on the masses would constrain @xmath42 . if we reduce the error on the end point of the spectrum in figure [ c5_mbbj ] to 20 gev which might be achievable with high luminosity running at lhc , the uncertainty on @xmath93 reduces to @xmath202 gev . as in the case of point 5 ,
the decay @xmath203 is allowed .
an analysis similar to that discussed there should be able to establish that this channel is open . in the case of these points , the negative @xmath109 solution can be eliminated .
the branching ratio for @xmath204 is shown in figure [ c1-branching ] as a function of @xmath91 for both signs of @xmath109 for the solutions in the allowed range .
it can be seen from this figure that the branching ratio is substantially larger for @xmath145 .
the decay chain @xmath205 leads to a small @xmath90 peak as shown in figure [ c1_mll ] which corresponds to @xmath206 .
the smallest branching ratio for the @xmath145 solution shown in figure [ c1-branching ] is 5.7% , approximately a factor of ten larger .
it is clear that these two cases can be distinguished in 10fb@xmath207 despite the fact thata larger data set may be required to measure the size of the peak shown in figure [ c1_mll ] . in the case of point 2 ,
the difference in branching ratios is even greater , again enabling elimination of the @xmath145 solution .
in this paper we have outlined a strategy that can be used to systematically explore supersymmetry assuming that it is discovered at the lhc .
we have given an example of a global variable ( @xmath8 ) that can be used to determine the mass scale of susy if nature has chosen the sugra model .
such global variables will be used to give the first indication of a signal independent of the type of susy model .
the production of heavy mass states virtually guarantees that events with very energetic jets will exist .
other susy models such as those with @xmath208parity violation @xcite may have no missing @xmath209 if the lsp decays within the detector .
if the lsp decays to leptons , then all susy events will have leptons and they can be used in a global variable .
if the lsp decays to hadrons , the jet multiplicity will rise and a variable similar to @xmath8 should be effective .
more detailed exploration will depend on the particular susy model . in the sugra models
, there is a general feature , namely that the second lightest neutralino almost always decays to @xmath210 if the channel is open and to @xmath211 with a substantial branching fraction if it is not . in the former case
, this will be the dominant source of @xmath13 and it will be discovered in this process via its decay to @xmath212 if it has not been seen at lep . in the latter case
the measurement of the position of the end point in the @xmath21 mass distribution provides a very precise measurement of the mass difference between two of the sparticles .
after first observing one of these signals , one will move up the decay chain to determine other quantities .
we have then illustrated , using specific examples , some techniques that can be used to determine masses and branching ratios of sparticles .
some of these quantities were then used to determine the fundamental parameters of the sugra model some of which can be determined with great precision .
the ultimate goal of such studies would be to use very many measurements to make an overconstrained fit to the model , rather in the same way that current data are used to test the standard model @xcite this work was begun during the summer study organized by the division of particles and fields of the american physical society .
we benefited from many discussions with attendees at that study and with colleagues on the atlas collaboration .
in particular we would like to thank stephan lammel , john bagger , alfred bartl , fabiola gianotti and daniel froidevaux .
the work was supported in part by the director , office of energy research , office of high energy physics , division of high energy physics of the u.s .
department of energy under contracts de ac0376sf00098 and de - ac02 - 76ch00016 .
accordingly , the u.s .
government retains a nonexclusive , royalty - free license to publish or reproduce the published form of this contribution , or allow others to do so , for u.s .
government purposes .
one of us ( js ) would like to thank the swedish national research council for support .
99 atlas collaboration , _ technical proposal _
, lhcc / p2 ( 1994 ) .
cms collaboration , _ technical proposal _ ,
lhcc / p1 ( 1994 ) .
h. baer , c .- h .
chen , f. paige , and x. tata , phys .
rev . * d52 * , 2746 ( 1995 ) ; phys
* d53 * , 6241 ( 1996 ) .
a. bartl , j. soderqvist _
et al . _ , to appear in _ proceedings of the 1996 dpf summer study _ ( snowmass , co , 1996 ) .
d. froidevaux and e. richter - was , presentation to atlas collaboration meeting ( june , 1996 ) .
nilles , phys . rep .
* 111 * , 1 ( 1984);h.e .
haber and g.l .
kane , phys .
rep . * 117 * , 75 ( 1985);l .
alvarez - gaume , j. polchinski and m.b .
wise , nucl . phys .
* b221 * , 495 ( 1983);l .
ibanez , phys .
118b * , 73 ( 1982);j.ellis , d.v .
nanopolous and k. tamvakis , phys
* 121b * , 123 ( 1983);k .
et al . _ prog .
phys . * 68 * , 927 ( 1982 ) .
f. paige and s. protopopescu , in _ supercollider physics _
d. soper ( world scientific , 1986);h .
baer , f. paige , s. protopopescu and x. tata , in _ proceedings of the workshop on physics at current accelerators and supercolliders _ , ed .
j. hewett , a. white and d. zeppenfeld , ( argonne national laboratory , 1993 ) . f. abe , _ et al .
rev . lett.*75 * , 11 ( 1995 ) .
s. abachi , _ et al .
_ fermilab - pub-96 - 177-e .
lett . * 73 * , 2667 ( 1994 ) , phys . rev.*d52 * 2605 ( 1995 ) .
s. abachi _
et al . _ ,
lett . * 74 * , 2632 ( 1995 )
. j. ellis , _ et al .
_ , nucl . phys . *
b238 * , 453 ( 1984);a .
gabutti _ et . al . , _ hep - ph/9602432;h .
baer and m. brhlik , phys . rev .
* d53 * , 597 ( 1996 ) . s. mrenna , hep - ph/9609360 .
l.j . hall and m.
suzuki , nucl.phys . *
b231 * , 419 ( 1984 ) .
see for example , d. schaile , in _ proc .
international conference on high energy physics ( ichep ) _ , ( glasgow , scotland , 1994 ) . | if supersymmetry exists at the electroweak scale , then it should be discovered at the lhc .
determining masses , of supersymmetric particles however , is more difficult . in this paper ,
methods are discussed to determine combinations of masses and of branching ratios precisely from experimentally observable distributions . in many cases such measurements alone
can greatly constrain the particular supersymmetric model and determine its parameters with an accuracy of a few percent .
most of the results shown correspond to one year of running at lhc at `` low luminosity '' , @xmath0 .
= 6.5 in = 9.0 in = -.5 in = -1 in # 1| # 1| = # 10= 0=0 1= 1=1 0>1 # 1 / # 1#2 = # 1 lbnl-39412 * precision susy measurements at lhc * * i. hinchliffe@xmath1 , f.e .
paige@xmath2 , m.d .
shapiro@xmath1 , j. sderqvist@xmath3 , and w. yao@xmath1 * @xmath1_lawrence berkeley national laboratory , berkeley , ca _ @xmath2_brookhaven national laboratory , upton , ny _
@xmath3_royal institute of technology ( kth ) , stockholm , sweden _ |
in this contribution i want to report on some recent developments in the area of multiloop feynman diagram calculations .
these developments touch widely separated areas of mathematics , ranging from knot theory to number theory @xcite as well as from combinatorical hopf algebras @xcite to the realm of noncommutative geometry @xcite and the classification of operator algebras . at the same time , on the physics side , they touch areas as separated as the ising model @xcite in three dimensions , the @xmath0-function in @xmath1 theory in four dimensions @xcite , and the recent identification of the unknown constant in the @xmath2-parameter of the standard model @xcite . in david
broadhurst s recent work you will find results concerning the identification of transcendental numbers arising from such finite parts of feynman diagrams with the hyperbolic covolume of hyperbolic knots @xcite .
such results , generalizing the by now familiar identification of knots , numbers and counterterms @xcite , point towards a deep and yet to be clarified connection between qft and ( hyperbolic ) geometry .
the recent results of andrei davydychev and bob delbourgo @xcite identifying one - loop triangles and boxes with the geometry of the hyperbolic tetrahedron point in the same direction .
all these results give testimony to the still premature and incomplete understanding of qft which we still have more than half a century after its discovery . on the other hand
they give hope that we begin to see some parts of the mathematics which underlies qft .
let me summarize the results achieved in recent years , and let me stress the most promising facts which deserve , in my opinion , future attention .
let us start with the consideration of a feynman diagram which has a non - vanishing superficial degree of divergence , but has no subdivergent graphs .
the coefficient of the divergence is then a well - defined number , and independent of the chosen regularization or renormalization scheme .
the study of such coefficients allows to observe some remarkable patterns : we can abstract from the nature of the particles realizing a given topology , and consider solely the topology of the graph as determined by internal propagators . whenever the topology of two different graphs matches , even in different theories , we expect the same transcendental number from these graphs . in fig.([f1 ] ) we see some graphs and their topology wich completely determines the transcendentals in their counterterms .
this lead to a knot - to - number dictionary .
fig.([f2 ] ) summarizes some of the knot and number classes identified so far @xcite .
hence we can classify topologies by assigning knots to feynman graphs @xcite , and find that graphs which deliver similar knots deliver similar transcendentals .
this pattern works well enough to abstract conjectures about the nature of these transcendental numbers @xcite .
so far severe numerical tests have confirmed these conjectures @xcite .
this allows to determine a search base of transcendental numbers in which a diagram will evaluate
. this knowledge suffices to turn a numerical answer for a diagram into an analytic one , as was demonstrated for the first time in @xcite .
the rational weights with which the transcedentals seen in counterterms contribute depend on the specific nature of the theory realizing a given topology .
it is a fascinating and still open problem to understand how these weights can be derived from the representation theory of the lorentz group .
meanwhile , we understand that the presence of extra symmetries like gauge symmetries @xcite or supersymmetry @xcite can annihilate the presence of certain transcendentals in the final result .
but then , if the above described primitive feynman diagrams have such wonders in store , what about general feynman diagrams , with full - fledged subdivergences ? there are clearly patterns in the study of subdivergent graphs which are hardly accessible with standard techniques of perturbation theory @xcite .
it was the question how to disentangle a general feynman graphs in terms of primitive graphs which lead to the discovery of a hopf algebra structure to which we now turn @xcite .
we change from the notion of parenthesized words , used in @xcite , to the notion of rooted trees .
this notion turned out to be very convenient in the recent thorough investigation of the mathematical role played by the hopf algebra of renormalization @xcite .
fig.([f3 ] ) summarizes some basic notions , as developped in @xcite and , with particular emphasis given to rooted trees , in @xcite .
fig.([f4 ] ) gives a diagrammatic explanation of this hopf algebra , and shows how the combinatorics of the forest formula derive from this hopf algebra .
the figure explains that the local counterterm appears as a fundamental operation in the hopf algebra .
it corresponds to the coinverse , the antipode .
it reveals that the calculus of renormalization can be completely understood as derived from an underlying hopf algebra structure .
this includes the case of overlapping divergences , which can be resolved in terms of this hopf algebra using either schwinger dyson equations @xcite , powercounting arguments and differential equations for bare green functions @xcite or algebraic methods @xcite .
the universal structure @xcite of the hopf algebra of @xcite gives hope that other diagrammatic expansions , for example asymptotic expansions , can be interpreted in terms of such algebraic structures as well .
let us also stress that the appearance of this hopf algebra structure in qft establishes a link to recent developments in mathematics .
quite the same hopf algebra turned up in the work of alain connes and henri moscovici @xcite . the precise relation is now clarified @xcite .
this gives a conceptual backing to the structure of local quantum field theories which ought to be thoroughly investigated in the future .
let us close these remarks with a simple observation which demonstrates the usefulness of the hopf algebra in the understanding of the appearance of transcendentals in feynman diagrams .
the results of @xcite point towards algebraic four - term relations between feynman diagrams . clearly , to fully understand all relations between counterterms of diagrams , some more work is needed .
one might hope to construct a basis of diagrams which have to be calculated , and then hopes to be able to determine all other diagrams by knowing relations between diagrams .
the simplest example of a four - term relation could appear at three loops , as at least two chords are necessary , by definition .
it is given in fig.([f5 ] ) .
relations between these four topologies are obscured by the fact that there will be subdivergences present . to overcome this problem we disentangle graphs into primitive elements of the hopf algebra of renormalization .
let us give an idea how this might help to achieve a full understanding of a ( modified ) four - term relation between diagrams .
we realize the four topologies in yukawa theory . assume
we expect that a four - term relation holds between the four graphs given in the figure , possibly modified by terms including four - point couplings .
such a modified relation is suggested by a detailed study of the results in @xcite .
these two extra terms , as well as two terms of the original four - term relation , are of ladder topology .
the two remaining terms have the @xmath3 topology , but a different structure as rooted trees .
now let us decompose the one involving the one - loop vertex correction as indicated in the figure .
this corresponds to a decomposition into primitive elements .
thus , for the four - term relation to hold , the only possible way is that the terms @xmath4 cancel on the level of antipodes .
this is indeed the case , as indicated in the figure , and will be reported in greater detail elsewhere . here
, it suffices to note that a decomposition into primitive elements of the hopf algebra agrees with a decomposition into the transcendental content of the various terms .
in this paper we succinctly reviewed a large body of results .
most of these results were gained through an investigation of feynman diagrams as entities of interest in their own right . altogether , the underlying idea that the numbers seen in the calculation of feynman diagrams are not that arbitrary , but on the contrary determined by their topology , underlies most of the results considered .
the relation to braid - positive knots comes as a surprise in the exploration of such ideas , and still awaits its final explanation .
the fact that there is a hopf algebra structure underlying renormalization theory points towards a beautiful connection between local quantum field theory and the theory of operator algebras , which , in modern terminology , is called non - commutative geometry .
if some of these ideas come to fruition in the future , we arrive at a picture of qft which is quite different from what we find in textbooks these days .
let us start with an arbitrary feynman graph .
we first construct its decomposition into primitive graphs , as determined by the hopf algebra structure .
these primitive graphs come as letters in which our field theory is formulated .
we determine a minimal basis in these letters , guided by relations between graphs as indicated by the study of a ( modified ) four - term relation .
the elements of this basis are feynman graphs which are connected to braid - positive knots , evaluating to counterterms expressed in transcendental numbers , which , in the best of all worlds , we could infer from the knowledge of a knot - to - number dictionary gained from empirical data or some deeper insight in the future .
this dictionary must not stop at the level of counterterms . the before mentioned results of david broadhurst on the finite part of diagrams @xcite indicate that the story continues .
if we then solve one further problem , how to determine the rational numbers which come as coefficients of these transcendentals , dependent on the spin representation of the involved particles , we approach an understanding of local qft which can be considered satisfactory .
being in an optimistic mode , one might even wonder if a decomposition in terms of primitives might have something to say about the asymptotic nature of the perturbative expansion ?
the perturbation series is asymptotic at best , but it might well be much better behaved if we disentangle it into rational contributions , contributions @xmath5 , and so on , having in mind that each of these numbers seems to refer to a different topology , which we might be forbidden to add up naively .
all of these properties seem to be restricted to a local point - like qft .
indeed , it seems that most of the successes attributed to theories based on extended objects are present due to the fact that these theories notoriously avoid the subtleties imposed on us by the presence of quantum fields localized at a point .
these subtleties are reflected by the presence of uv - divergent integrals .
but then a careful study of the properties of these uv divergences indicate that they encapsulate an enormously rich structure at high loop orders , pointing towards the mathematics which might be needed to extend our understanding of qft in the future .
it is a pleasure to thank the organizers for a lively and stimulating workshop . as always
, i have to thank david broadhurst for an uncountable number of discussions and all his enthusiasm , and i thank alain connes for guiding me into the realm of noncommutative geometry and for lot of illuminating discussions .
i thank the dfg for a support by a heisenberg fellowship .
99 d. kreimer , j.knot th.ram.*6 * , 479 ( 1997 ) , q - alg/9607022 .
d. kreimer , adv.math.theor.phys.*2 * , 71 ( 1998 ) , q - alg/9707029 .
a. connes , _ noncommutative geometry _ , academic press , 1994 .
a. connes , h. moscovici , _ hopf algebras , cyclic cohomology and the transverse index theorem _ , ihes / m/98/37 .
broadhust , _ a dilogarithmic 3-dimensional ising tetrahedron _ , hep - th/980525 .
broahurst , d. kreimer , int.j.of mod.phys.*c6 * , 519 ( 1995 ) .
broadhurst , _ massive 3-loop feynman diagrams reducible to sc * primitives of algebras of the sixth root of unity _
, hep - th/9803091 .
broadhurst , _ solving differential equations for 3-loop diagrams : relation to hyperbolic geometry and knot theory _ , hep - th/9806174 .
broadhurst , d. kreimer , phys.lett.*b393 * , 403 ( 1997 ) .
broadhurst , j.a .
gracey , d. kreimer , z.phys.*c75 * , 559 ( 1997 ) .
broadhurst , r. delbourgo , d. kreimer , phys.lett.*b366 * , 421 ( 1996 ) .
d. kreimer , phys.lett.*b354 * , 117 ( 1995 ) .
d. kreimer , _ knots and feynman diagrams _ , cambridge up , in press .
a. davydychev , r. delbourgo , _ a geometrical angle on feynman integrals _ hep - th/9709216 , to appear in j.math.phys . ;
+ a. davydychev , r. delbourgo , _ geometrical approach to the evaluation of multileg feynman diagrams _
, hep - th/9806248 .
broadhurst , _ conjectured enumeration of irreducible multiple zeta values , from knots and feynman diagrams _
, to appear in phys.lett.b , hep - th/9612012 .
borwein , d.a .
bradley , d.j .
broadhurst , electronic j. combinatorics * 4 * ( 1997 ) r5 .
p.m. ferreira , j.a .
gracey , phys.lett.*b424 * , 85 - 92 ( 1998 ) , hep - th/9712140 .
d. kreimer , eur.phys.j.*c2 * , 757 ( 1998 ) , hep - th/9610128 .
a. connes , d. kreimer , _ renormalization , hopf algebras and noncommutative geometry _ , in preparation .
d. kreimer , _ on overlapping divergences _ , in preparation .
d. kreimer , j.knot th.ram.*7 * , 61 ( 1998 ) , hep - th/9612010 .
broadhurst , d. kreimer , phys.lett.*b376 * , 329 ( 1998 ) , hep - th/9612011 . | we summarize recent results connecting multiloop feynman diagram calculations to different parts of mathematics , with special attention given to the hopf algebra structure of renormalization .
pacs : 11.10gh , 11.15bt , 11.10-z , 02.90+p + |
attempts to characterize the distribution of mass , orbital radius , and eccentricity of extrasolar planets are hampered by the lack of knowledge of detection sensitivities .
following our previous study of the lick survey ( cumming , marcy , & butler 1999 , hereafter cmb ) , the aim of this work is to carefully assess the detection threshold and selection biases for a sample of stars from the keck survey , and use this information to constrain the distribution of orbital parameters .
we analyse radial velocity measurements for 580 f , g , k , m stars from the keck survey .
typically , each star has 3 or 4 measurements per year over 2 to 5 years with measurement errors between 2 and 5 m / s . we adopt a least - squares approach .
first , we check whether the velocity scatter is consistent with measurement errors and intrinsic stellar `` jitter '' .
the jitter , which arises from a combination of convective motions , magnetic activity , and rotation , is estimated from stellar parameters such as rotation period and mass .
assuming gaussian statistics , we calculate the probability of obtaining @xmath0 as large or larger than the observed value from measurement errors and jitter alone , adopting a threshold probability of @xmath1 .
the solid line in figure 1 shows the signal to noise needed for a signal to be detected by this method 50% of the time as a function of the number of data points @xmath2 .
we define the signal to noise as @xmath3 , where @xmath4 is the velocity amplitude , and @xmath5 the expected noise amplitude . even with only a few measurements , signals @xmath6 are detected .
next , we look for periodicities using the lomb - scargle periodogram ( scargle 1982 ; walker et al . 1995 ;
cmb ) , which is similar to the f - statistic , measuring the improvement in @xmath0 between a model which includes a sinusoid and one that does not .
we look for the maximum periodogram power between 2 days and 10 years , and use monte carlo methods to assess its significance .
the dashed lines ( analytic = 3.75 in result ) and squares ( numerical result ) in figure 1 show the signal to noise ratio needed to identify a periodic signal 50% ( filled squares ) and 90% ( open squares ) of the time ( significance level @xmath1 ) as a function of number of data points . to obtain analytic estimates ,
we follow the approach of horne & baliunas ( 1986 ) , but using the correct statistical distribution for periodogram powers in the presence of noise ( cmb appendix b ) .
we find the signal to noise ratio needed for detection 50% of the time is @xmath7^{1/2}$ ] , where @xmath8 the number of independent frequencies searched , and @xmath9 the false alarm probability needed for detection . for large @xmath2
the 50% threshold is @xmath10^{1/2}$ ] , or @xmath11 for typical values @xmath12 and @xmath13 .
figure 1 is striking because when looking for periodic signals , we are used to being able to `` dig into the noise '' .
however , here we are limited by the small number of observations .
points to note are : ( i ) even with small @xmath2 , we detect excess variability if @xmath14@xmath15 , ( ii ) it is hard to characterize an orbit when the number of observations is @xmath16 , ( iii ) we detect periodic signals with @xmath17@xmath18 when @xmath19@xmath20 , and ( iv ) detecting amplitudes @xmath21 requires @xmath22 .
we are currently extending this approach to include eccentric orbits .
however for now , we use the best - fit circular orbit amplitude and period from the periodogram as initial conditions for a full keplerian fit . the resulting keplerian amplitudes and periods for the significant detections
are shown in figure 2 .
most , but not all , of the candidates from the periodogram analysis are announced planets ( open circles ) . those candidates that have not been announced ( filled squares ) have either low amplitude or long orbital period . there is an `` effective detection threshold '' beyond that set by the statistics of the periodogram .
this comes from ( i ) we must be sure the observed variations are not due to periodic stellar jitter , and ( ii ) inability to characterize orbits longer than the duration of the observations .
the announced planets have velocity amplitude @xmath23@xmath24 , periods less than the duration of the observations ( @xmath25@xmath26 ) , and number of data points @xmath27 .
better characterization of jitter would aid detection of low amplitude planets .
finally , for most stars with no detection , we can exclude companions with velocity amplitudes @xmath23@xmath28 , or @xmath29@xmath30 , for orbital periods less than the duration of the observations ( 2 to 5 years ) .
we adopt a bayesian approach to constrain the distribution of masses , orbital periods , and eccentricities _ using both detections and upper limits_. these calculations are currently in progress , here we discuss the methodology .
consider a model of the distribution of masses and orbital periods @xmath31 ( normalized so that @xmath32 ) and fraction of stars with planets @xmath33 .
we write a likelihood function @xmath34 where @xmath35 denotes each star , and @xmath36 this expression includes the probability of the data _ given a planet _ of mass @xmath8 , period @xmath37 , @xmath38 ^ 2\over 2\sigma_k^2}\right),\ ] ] and the probability of the data _ given no planet _ @xmath39 .
no decision must be made for each star about whether a planet has been detected or not : both possibilities are included , weighted by their relative probability .
this approach can be readily generalized to include , for example , eccentric orbits . as a check ,
imagine we can definitely say whether a star has a planet or not . the likelihood is
then @xmath40 where @xmath41 planets are detected out of @xmath42 stars .
maximizing @xmath43 with respect to @xmath33 gives the expected result @xmath44 , the best guess at the fraction of stars with planets is the observed fraction .
two striking features already apparent are ( i ) the lack of massive planets at short periods ( easiest to detect , but not seen ) , and ( ii ) the `` pile - up '' of planets at orbital periods @xmath45 ( no planets were found between @xmath46 and @xmath47 days in our search , suggesting that this is not a selection effect ) .
we are currently investigating these issues and others more carefully , including ( i ) the fraction of stars with planets , ( ii ) whether there is a lack of saturn - mass companions at short orbital periods , and ( iii ) whether planet properties depend on the spectral type of the host star .
we thank g. ushomirsky , d. chernoff , i. wasserman , r. rutledge and d. reichart for useful discussions .
ac thanks caltech astronomy department for hospitality during a recent visit . this work was supported by nasa hubble fellowship grant hf-01138 awarded by the space telescope science institute , which is operated for nasa by the association of universities for research in astronomy , inc .
, under contract nas 5 - 26555 . | we present an analysis of precision radial velocity measurements for 580 stars from the keck survey .
we first discuss the detection threshold of the survey , and then describe a bayesian approach to constrain the distribution of extrasolar planet orbital parameters using both detections and upper limits . |
evidences have accumulated in recent years that the peak effect ( pe ) phenomenon in critical current density ( @xmath9 ) located at the edge of the depinning process is distinct from the second magnetization peak ( smp ) anomaly commencing at field values deeper in the mixed state of the weakly pinned superconducting samples of several low @xmath10 and high @xmath10 compounds , e.g. , ca@xmath3rh@xmath4sn@xmath5 @xcite , rni@xmath1b@xmath1c ( @xmath11 ) @xcite , v@xmath3si @xcite , @xmath0-nbse@xmath1 @xcite , pros@xmath4sb@xmath12 @xcite , yba@xmath1cu@xmath3o@xmath13 @xcite , etc . in collective pinning @xcite scenarios ,
the critical current density can be related to the length scales over which the displacements of the flux lines from their equilibrium positions remain correlated . in such frameworks ,
an anomalous increase in field / temperature ( h / t ) variation of @xmath9 can be construed to imply a decrease in the correlation volume over which flux lines remain elastically deformed and collectively pinned , thereby , leading to the notion of order - disorder transformation in the underlying flux line lattice ( fll ) @xcite .
the occurrence of two anomalous variations in @xmath14 ( e.g. , the smp and the pe ) in a given isothermal scan may , therefore , imply that the disordering of the ordered lattice can proceed in two distinct steps , before reaching the depinning limit in the bulk of the sample @xcite . a notion of weak
pinning by surface inhomogeneities may survive , after crossing the bulk depinning limit , upto the upper critical field ( @xmath15 ) @xcite .
recent studies of the metastability effects @xcite in single crystals of @xmath0-nbse@xmath1 , which display a smp like anomaly prior to pe , have revealed that ordered and disordered regions co - exist from above the onset position of smp anomaly upto a limiting temperature , designated as the spinodal temperature ( @xmath16 ) , which lies above the peak position of pe @xcite . a spinodal line is the locus of @xmath17 above which the path dependence in @xmath18 is absent @xcite .
the notion of spinodal line is rationalized by terming the order - disorder transition in fll of @xmath0-nbse@xmath1 system to have a first order character , in analogy to the fll melting line in the weakly pinned crystals of high @xmath10 bismuth cuprates @xcite .
li and rosenstein @xcite have theroetically argued that metastability associated with the order - disorder first order transition line could span a large part of the ( @xmath19 ) phase space , starting from the spinodal line and extending sufficiently below the boundary marking the onset position of the order - disorder transition .
using fast current ramp procedures in transport measurements , xiao _ et al _ @xcite had exemplified the notion of a generic spinodal line via a normalized field - temperature ( @xmath20 ) plots in a number of crystals of @xmath0-nbse@xmath1 , which are in agreement with the theoretical results of li - rosenstein @xcite .
the single crystal samples studied by xiao _
et al _ @xcite displayed only the pe phenomenon . extending these studies , li _
et al _ @xcite have recently exploited the time resolved transport measurements in a high purity single crystal of @xmath0-nbse@xmath1 ( @xmath21 ) to elucidate the marking of a limiting temperature @xmath22 far below the onset temperature of pe ( @xmath23 ) , below which ( i.e. , @xmath24 ) , fll is essentially dislocation free . above @xmath22 , a sharp change in the reordering timescales , observation of strong hysteresis , and the presence of a plastic response , reflects the proliferation of dislocations , which they surmise as a precursor of the pe transition to the disordered state . these results in turn imply that in the ( h , t ) phase - space between the onset of precursor effects and upto the spinodal line ( i.e. , between @xmath22 and @xmath16 lines ) , the manifestation of tendency to polycrystallinity in the vortex matter can be explored in a variety of vortex dynamics measurements .
explorations by banerjee _
et al _ @xcite and thakur _ et al _ @xcite in a variety of crystals of @xmath0-nbse@xmath1 reveal an intimate connection between hysteretic effects ( i.e. , metastability / thermomagnetic response ) and effective pinning , such that an increase in quenched random disorder leads to surfacing of a smp anomaly prior to the pe phenomenon .
the results of li _ et al _ @xcite have motivated us to explore a manifestation of the evolution in the reordering time scales when the notion of phase co - existence @xcite spans across the regions of smp and pe anomalies @xcite , and it can be bifurcated into two sub - parts , where ordered and disordered pockets dominate , respectively . in particular , in the sub - part between the onset positions of smp and pe , where the ordered pockets dominate , there is a possibility of further sub - division . while the increase in @xmath9 from the onset position of smp ( @xmath25 ) anomaly to its peak position ( @xmath7 ) can be termed as anomalous , the decrease in @xmath9 from @xmath7 to the onset position of pe ( @xmath26 ) is indeed non - anomalous @xcite .
this permits a non - monotonic variation in the relative weights of ordered and disordered pockets in the phase co - existence regime between @xmath25 and @xmath8 .
such a behavior could get reflected in the reordering time scales experienced during isothermal magnetization hysteresis experiments , where the magnetic field is steadily ramped as in a vibrating sample magnetometer ( vsm ) . in field ramping experiments , disordered vortices enter a given sample through its edges / surface inhomogeneities @xcite and they attempt to reach towards an underlying equilibrium state in specific experimental conditions @xcite .
we present here an analysis of the results of one such study performed using a vsm .
we believe that new data are supportive of a claim @xcite that two distinct features ( viz . ,
the smp and pe anomalies ) evident in the disordering of the weakly pinned fll represent two distinct phase transitions .
the single crystals of @xmath0-nbse@xmath1 used in this study are , ( i ) a nascent pinned sample @xmath27 ( @xmath28 ) with @xmath29 and ( ii ) few somewhat more strongly pinned samples , viz . , @xmath30 ( @xmath31 ) and @xmath32 ( @xmath33 ) , having 200 ppm of fe as impurity and with @xmath34 .
the residual resistivity ratio ( @xmath35 ) for the samples @xmath27 and @xmath32 were measured to be 20 and 9.8 , respectively .
the crystal of ca@xmath3rh@xmath4sn@xmath5 ( @xmath6 ) is the same @xcite as utilized by sarkar _
et al _ @xcite .
isothermal m - h loops were recorded on a 12 tesla vibrating sample magnetometer ( vsm ) ( oxford instruments , uk ) , with field scan rates ( @xmath36 ) in the range of 100 oe / min to 8 koe / min .
the ac susceptibility measurements were done utilizing an ac option facility on a commercial squid magnetometer ( quantum design inc .
, u.s.a . , model mpms7 ) .
1 ( a ) shows a portion of an isothermal m - h loop recorded at 3.86 k in crystal @xmath32 of @xmath0-nbse@xmath1 , with the magnetic field having being ramped with @xmath37 of 3 koe / min . the pe and smp like anomaly have been identified , with the boxed region showing the latter , which sits deeper in the mixed state .
the same figure also depicts magnetization values ( open triangles ) at different fields obtained in field cooled ( fc ) manner ( @xmath38 ) . in view of the tendancy
that notion of pinning resists the expulsion of flux from the interior of the sample during field cooling process , the small angle neutron scattering experiments to probe the structure of the vortex matter at a given field are performed in the field cooled conditions @xcite . in a fc state ,
the magnetization value can , therefore , be taken to represent a notional equilibrium magnetization value ( @xmath39 ) , with no gradient in macroscopic field in the bulk of the sample .
the width of the magnetization hysteresis loop ( @xmath40 ) at a given field is usually considered to measure the critical current density @xmath14 @xcite . the hysteresis width is observed to depend on the scan rate @xmath37 of the field .
the linear relationship between @xmath40 and @xmath14 is satisfactory when field fully penetrates the sample and its variation over the sample is not large . in the given crystal @xmath32 ,
the full penetration field at 3.86 k in the field range of interest is less than 100 oe @xcite .
we may , therefore , assume that in the field interval of interest ( 1 koe to 15 koe ) in the present experiment in sample @xmath32 at 3.86 k , the variation in the local macroscopic field across the sample is a very small fraction of the externally applied field .
if we , therefore , assume that the applied field value ( notionally ) prevails across the sample , it will amount to a little quantitative inaccuracy . in fig . 1 ( b )
, we have shown the effect of @xmath37 on one - half of the hysteresis width , @xmath41 ( @xmath42 and proportional to the critical current density , @xmath43 , during the forward leg of the envelope curve from 1 koe to 15 koe .
these data suggest a substantially larger scan rate dependence of the @xmath43 values prior to the pe region .
it is apparent that @xmath14 at any h decreases as ramp rate decreases .
an instantaneous value of @xmath14 would correspond to the limit of an infinite scan rate , i.e. , @xmath44 .
slower ramping of field permits larger settling time for injected disordered bundles of vortices to approach towards an underlying equilibrium state at a given @xmath45 . in order to elucidate the modulation in the field dependence of @xmath14 vs @xmath37 before and after the order - disorder transition(s ) as evidenced by smp and pe
, we plotted @xmath46 vs @xmath47 at different values of @xmath45 . from every such plot ,
we extracted by extrapolation a limiting magnetization value ( @xmath48 ) for an infinite scan rate ( @xmath49 ) . using these values , we show in the panels ( a ) and ( b ) of fig .
2 the log - log plots of @xmath50 ( proportional to the normalized critical current density , @xmath51 ) versus @xmath47 at different field values , starting from deep within the mixed state , right upto the peak position of the pe at 3.86 k. from these data , it is evident ( see fig .
2 ( a ) ) that the temporal decay of the forward critical current density progressively increase as one proceeds from the low field end ( e.g. , 1.6 koe ) , till one approaches the onset field of the smp , @xmath25 ( @xmath52 ) . above that field
, there is a relative decrease in the temporal decay , until the peak position of smp , @xmath7 ( @xmath53 ) is reached .
2 ( b ) shows that from @xmath7 to the onset field for the pe , @xmath8 ( @xmath54 ) , there is again an increase in the temporal decay . beyond @xmath8 ,
the temporal decay reduces monotonically upto the peak position of the pe , @xmath55 ( @xmath56 ) .
these observations , we believe , attest to the presence of multiple undulations in the reordering time scales of the injected disordered vortices . in order to further comprehend the modulation in the temporal decays in @xmath51 ,
we have used a parameter @xmath57 defined as @xmath58 , which we believe , represents the normalized decay rate of @xmath9 at larger times ( i.e. , very slow scan rate ) .
the inset and main panels of fig .
3 show behavior of @xmath59 at 3.86 k in the sample @xmath32 of @xmath0-nbse@xmath1 .
we indeed observe a non - monotonic variation in @xmath60 , until it reaches a limit near @xmath61 , which we reckon correspond to a value that such a parameter would assume for the usual thermal creep behavior @xcite in the pinned state .
as stated earlier , the observed behavior can be rationalized on the premise that if the underlying equilibrium state at a given h is completely disordered , the injected disordered bundles of vortices would assume such a state instantaneously and , therefore , the time decay ( i.e. , scan rate dependence ) in hysteresis width ( @xmath62 ) would be governed by the process of thermal creep alone . however ,
if the underlying equilibrium state is partially ordered , the injected vortices would require additional time lapse to conform to that state before the process of thermal creep starts governing the observed time decay .
in an isothermal ramp , at low fields ( where fll lattice constant @xmath63 , the penetration depth ) , the injected diordered vortices could be expected to first conform to the reentrant disordered state @xcite , followed by ( gradual ) crossover to the progressively better ordered elastic glass phase @xcite upto the onset of the smp anomaly .
thereafter , as the state of co - existence of weaker and stronger pinned phases commences @xcite , the injected disorder would attempt to heal towards the partially ordered equilibrium state , which can be attained in somewhat shorter time scales than that needed for @xmath64 .
we believe that the modulation in @xmath60 in fig .
3 essentially reflect the changes in the underlying equilibrium state from reentrant disordered phase to the peak position of pe .
we are tempted to attribute the turnaround in @xmath60 response between @xmath7 and @xmath8 to the possibility of improvement in the state of partial order with increase in field in this interval @xcite . considering that the @xmath60 value at @xmath8 is about the same as that at @xmath25
, we may naively state that the partially disordered phase nearly heals uptill the arrival of the pe regime .
we show in fig .
4 the scan rate dependence of the ratios of the critical current density at the peak position to that at the onset position for the smp and pe anomalies , respectively obtained from the isothermal m - h data at 3.86 k. the increase in the values of the ratio with decrease in scan rate conforms to the notion that slower ramp rate permits attainment of superior spatial order prior to entry into the anomalous regimes . for the smp anomaly , the ratio changes from about 1.3 to about 2.5 , whereas for the pe , it changes from about 20 to about 90 .
let us recall that in a collective pinning scenario due to larkin - ovchinnikov @xcite , @xmath9 relates inversely to the correlation volume , @xmath65 ( @xmath66 , where @xmath67 and @xmath68 are radial and longitudinal correlation lengths ; for simplicity we may take @xmath69 ) . the above values then imply a reduction in radial correlation length ( @xmath67 ) by a factor of about 20 across the pe .
et al _ @xcite had estimated @xmath70 ( where , @xmath71 is fll constant ) at @xmath72 to be @xmath7320 at 4.2 k in a crystal ( sample @xmath30 ) of @xmath0-nbse@xmath1 ( @xmath34 ) .
a change in @xmath67 value by a factor of about 20 during complete amorphization of fll between the onset and peak positions of pe therefore appears satisfactory .
when both smp and pe are present in a given isothermal scan , the collapse in correlation volume of vortex lattice could occur in two stages - a very large domain first sub - divides into multiple parts , followed by partial healing and ( eventual ) complete amorphization into micro - domains . to establish the efficacy of prescription of analysis of scan rate dependence of hysteresis width as illustrated in fig . 2 and fig .
we show for comparison in fig .
5 the @xmath60 data at two temperatures in a weakly pinned crystal of another low @xmath10 superconductor , viz . , ca@xmath3rh@xmath4sn@xmath5 , in which smp and pe had been reported earlier @xcite . the smp anomaly in this compound is as prominent as the pe , the field region of smp , however , is insensitive to the temperature variation , whereas the pe closely follows the @xmath15 line .
the separation between the smp anomaly and the pe , therefore , decreases as @xmath74 increases .
it had also been suggested that in ca@xmath3rh@xmath4sn@xmath5 , the dislocations injected in between the onset and peak positions of smp could heal upto the onset position of the pe @xcite .
it is useful to examine the @xmath60 behavior in ca@xmath3rh@xmath4sn@xmath5 at 1.9 k and 3.6 k ( cf .
5 ( a ) and 5 ( b ) ) .
while the non - monotonic behavior in @xmath60 is evident at both temperatures , the relative enhancement in @xmath60 values between @xmath25 and @xmath7 is more at 1.9 k as compared to that at 3.6 k. considering the larger field interval between @xmath25 and @xmath7 in ca@xmath3rh@xmath4sn@xmath5 as compared to the corresponding interval in @xmath0-nbse@xmath1 , the @xmath60 values decrease to a much deeper level in the former sample , and from which the turnaround also happens upto a lower limit , i.e. , @xmath75 value is smaller than @xmath76 value in ca@xmath3rh@xmath4sn@xmath5 .
a comparative look at @xmath60 data in crystal @xmath32 of @xmath0-nbse@xmath1 and that of ca@xmath3rh@xmath4sn@xmath5 conveys similarity in behavior across smp and pe anomalies ( cf .
3 and fig .
5 ) , however , the values of the @xmath57 parameter in the former case are much larger .
the observation that smp anomaly in ca@xmath3rh@xmath4sn@xmath5 does not display temperature variation is supportive of the viewpoint that this anomaly is entirely pinning induced .
a study of the variation of quenched random pinning on the pe phenomenon in a number of crystals of @xmath0-nbse@xmath1 had independently revealed that smp anomaly in this system surfaces up with the enhancement of pinning .
it is , therefore , instructive to compare the scan rate dependences of hysteresis width in crystals of @xmath0-nbse@xmath1 , with varying pinning strengths . in a given crystal of @xmath0-nbse@xmath1 , effective pinning can also be seen to enhance with the increase in field @xcite .
we present in fig .
6 glimpses into few above stated results obtained in crystals @xmath32 and @xmath30 ( both having @xmath34 ) and a nascent pinned crystal @xmath27 of @xmath0-nbse@xmath1 ( @xmath21 ) .
( a ) shows plots of critical current density on the forward leg normalized to its value at @xmath77 , @xmath78 ( @xmath79 ) , at scan rates of 3 koe / min and 0.3 koe / min , respectively , at a temperature of 3.07 k in the crystal @xmath32 of @xmath0-nbse@xmath1 .
these data correspond to a situation , where the smp anomaly partially overlaps with the pe anomaly ( broadening due to enhancement in effective disorder ) and one can barely mark the onset position @xmath8 of the pe . even in such a circumstance , one can see a marked difference in the dynamical response above the onset field of the pe anomaly and at field values below it .
one may be tempted to ignore the distinctiveness of the region of smp anomaly on the basis of data in fig . 6 ( a ) .
an inset panel in fig .
6(a ) shows the ramp rate dependence of normalized @xmath40 at 2 k in the crystal @xmath30 of @xmath0-nbse@xmath1 .
here , one can notice that at a slower scan rate of 0.25 koe / min .
, the signature corresponding to the smp like anomaly becomes indiscernible , while it is distinctly observable at a larger scan rate of 8 koe / min .
such an observation could lead to an apprehension whether any change occurs in the underlying vortex state across the field interval of @xmath25 and @xmath7 . in this context
, we now draw attention to the scan rate dependence of the hysteresis width across pe in crystal @xmath27 ( see fig . 6 ( b ) ) . an inset panel in fig .
6(b ) shows a comparison of the forward legs of two m - h loops obtained with @xmath80 and @xmath81 , respectively in crystal @xmath27 at 2 k. it is apparent that the hysteresis width across pe region shrinks rapidly with decrease in @xmath37 .
attempts to obtain a hysteresis bubble in the m - h data across pe recorded using a squid magnetometer ( quantum design inc . u.s.a . ) , where data are recorded in a quasi - static manner ( after a time lapse of about a minute from the instance the magnet current is set in persistent mode ) did not yield affirmative output , as the hysteresis bubble had collapsed to the background limit in the elapsed time @xcite . non - observation of pe bubble in m - h data in very clean crystals of @xmath0-nbse@xmath1 has been eluded to earlier @xcite , however , the changes ( amounting to a phase transition ) happening in the vortex state across the pe have been elucidated via stm imaging in the same crystal @xcite .
we are , therefore , inclined to persue probing the field interval of smp region in the crystal @xmath32 of @xmath0-nbse@xmath1 . in fig . 7 , we show plots of the widths of hysteresis bubble at @xmath82 and @xmath61 normalized to its extrapolated value at @xmath83 , as a function of @xmath84 at 3.86 k in crystal @xmath32
we have also included in this figure the plot of the hysteresis width at @xmath61 in crystal @xmath27 at 2 k. note the similarity in the behavior of plot pertaining to pe in crystal @xmath27 with that pertaining to smp in crystal @xmath32 .
both the hysteresis widths rapidly decrease with increase in @xmath84 , thereby , implying the difficulty in their observation in the static limit while recording data with a squid magnetometer @xcite .
the width of the pe bubble in crystal @xmath32 , however , decreases slowly with increase in @xmath84 and this anomaly , therefore , registers its presence in the data recorded with the squid magnetometer .
the observations in fig . 6 and fig .
7 prompt us to state that enhancement in pining in crystals of @xmath0-nbse@xmath1 not only invokes the surfacing of smp , but , also slows down the time decay of macroscopic currents that get set up in response to a driving force in the pe region .
one difference between smp and pe anomalies in ca@xmath3rh@xmath4sn@xmath5 and the sample @xmath32 of @xmath0-nbse@xmath1 is that in the former both the features are well separated at lower temperatures ( higher fields ) , but in the latter sample , smp appears as a precursor part of pe , with the possibility of large overlap at lower temperatures . when smp and pe lie in closer proximity , the two together present themselves as a candidate @xcite of edge effects complicating the pe pertaining to order - disorder transition in the bulk of the sample in ac susceptibility measurements . in order to further explore the case of smp as a bulk feature and/or edge effects , we measured ac susceptibility in sample @xmath32 of @xmath0-nbse@xmath1 at different frequencies and amplitudes of ac field .
fig . 8 ( a ) shows isofield ( @xmath85 ) ac susceptibility ( @xmath86 ) data measured with an @xmath87 of 2.5 oe ( r.m.s . ) in the frequency range 1 hz to 1111 hz .
it can be noted that the signatures corresponding to smp and pe anomalies remain distinctly observable over the entire range of frequencies , thereby discounting the possibility of smp anomaly as being an edge effect .
8 ( b ) shows the @xmath86 data measured with different amplitudes of the ac drive ( @xmath87 ) at a frequency of 111 hz . the omnipresence of the two anomalous variations is self evident in fig . 8 .
change of frequency by three orders ( @xmath88 ) is expected to compromise the edge effects in a preferential manner @xcite , however , this does not seem to be the case in the present data . we surmise that smp and pe represent occurrence of two distinct order - disorder transitions in ( h , t ) phase space , whose locations can be marked from appropriate sets of data .
9 shows the vortex phase diagram for h@xmath89c in crystal @xmath32 of @xmath0-nbse@xmath1 , sketched from m - h data recorded at a high scan rate ( 3 koe / min ) .
the reentrant disordered phase , the bragg glass ( bg ) phase , the phase co - existence region ( vortex glass ( vg ) phase ) and the surface pinning region have been identified .
the bifurcation of the phase co - existence region into regions i and ii , where ordered and disordered pockets of vortices dominate , respectively , is also shown .
this diagram covers a larger portion of ( h , t ) space as compared to the one presented earlier @xcite in this sample .
it is appropriate to add here a caveat that phase boundaries in fig .
9 have been drawn from bulk magnetization studies , these would get somewhat modified , when drawn using local magnetization measurements . the construction of the vortex phase diagram in fig .
9 presumes the applicability of notions of bragg glass to vortex glass transition and an inverse relationship between the critical current density and the spatial extent over which the vortices remain correlated in the spirit of larkin - ovchinnikov collective pinning theory @xcite . by combining electrical transport studies in strip and corbino geometry with bitter decoration patterns at low fields ( viz . ,
36 oe , where intervortex spacing @xmath90 penetration depth @xmath91 ) in @xmath0-nbse@xmath1 for @xmath92 , it has been surmised @xcite that no obvious correlation exists between the topology of the vortex structure and enhancement in critical current in the ( partially ) ordered ( i.e. , polycrystalline ) vortex matter .
an extension of the notion of absence of any correspondence between the critical current density and the spatial order in the vortex state in the field region where interaction effects between vortices are appreciable ( @xmath93 ) could , in principle , undermine the viewpoint pursued above .
we have presented experimental data from dc magnetization measurements to elucidate the notion of a modulation in the amount of the ordered ( disordered ) fraction in the underlying equilibrium / steady state in a field ramp up cycle in weakly pinned samples showing second magnetization peak anomaly and peak effect phenomenon .
the annealing to an underlying equilibrium / steady state of the transient disordered vortex bundles injected into the sample during the field ramp in a vsm measurement has been used as a prescription to gain information on the extent of order / disorder prevailing in the vortex state at a given ( h , t ) . from an analysis of the ramp rate dependence of magnetization hysteresis width , a parameter @xmath57 ( @xmath94 ) has been projected , which purports to correlate to the time required to assume an underlying equilibrium state .
observation of non - monotonic variation in @xmath60 can be understood within a premise that the extent of spatial order in the equilibrium / steady state of the vortex matter at a given @xmath45 determines the ( initial ) scan rate dependence of the hysteretic magnetization response .
the magnetization value at a given instance at a particular field during a field ramp cycle carries information on the incomplete annealing of the stronger pinned disordered pockets towards the underlying equilibrium / steady state @xcite .
when the underlying equilibrium / steady state is anticipated to have a high degree of order , the injected disordered bundles require longer time lapse and resultant @xmath60 value is large . on the other extreme , when the destined equilibrium / steady state is disordered in nature ,
the injected disordered bundles assume that state instantly and s parameter is small ( its value corresponding to flux creep alone ) . at the low field end , where vortices are far apart and individually pinned , the vortex state is disordered . in weakly pinned samples showing the pe phenomenon , vortex solid is disordered in equilibrium above the peak position of pe .
modulations in s as a function of field between @xmath95 and @xmath55 reflect the quality of spatial order in the intervening region .
in particular the relative weights of the ordered / disordered pockets across the phase co - existence regime manifest themselves in a spectacular manner . an increase in the value of s between @xmath96 and @xmath8 hints towards an improvement in the state of spatial order in this field interval due to possibility of squeezing out of some of the dislocations injected into the sample at the onset position of the smp anomaly .
we would like to acknowledge numerous discussions with s. s. banerjee , a. a. tulapurkar and ms .
d. jaiswal - nagar .
we would also like to thank shobo bhattacharya for his comments .
we are grateful to c. v. tomy for the crystal of ca@xmath3rh@xmath4sn@xmath5 grown at warwick university , u.k .
one of us ( adt ) would like to acknowledge the tifr endowment fund for the kanwal rekhi career development fellowship .
d.jaiswal-nagar , d. pal , m. r. eskildsen , p. c. canfield , h. takeya , s. ramakrishnan , and a. k. grover , current trends in vortex state physics , editors : s. k. malik , a. k. grover , and s. n. bhatia , pramana - j .
phys . * 66 * , 113 ( 2006 ) .
a. d. thakur , s. s. banerjee , m. j. higgins , s. ramakrishnan , and a. k. grover , current trends in vortex state physics , editors : s. k. malik , a. k. grover , and s. n. bhatia , pramana - j .
phys . * 66 * , 159 ( 2006 ) .
s. s. banerjee , s. ramakrishnan , a. k. grover , g. ravikumar , p. k. mishra , v. c. sahni , c. v. tomy , g. balakrishnan , d. mck .
paul , p. l. gammel , d. j. bishop , e. bucher , m. j. higgins , and s. bhattacharya , phys .
b * 62 * , 11838 ( 2000 ) . ) at 3.86 k measured at a field scan rate of 6 koe / min in the crystal @xmath32 of @xmath0-nbse@xmath1 .
the field cooled magnetization ( @xmath38 ) data points are also plotted as open triangles , and the pe and the smp regions have been marked .
panel ( b ) shows plots of @xmath46 ( obtained from the forward leg of the envelope m - h loop ) at field scan rates of 0.1 koe / min , 0.3 koe / min and 6 koe / min , respectively . ] as a function of field at scan rates of 0.3 koe / min and 3 koe / min for the sample @xmath32 of @xmath0-nbse@xmath1 at 3.07 k. an inset in panel ( a ) shows the magnetization hysteresis width in the sample @xmath30 of @xmath0-nbse@xmath1 at 2 k as the scan rate is varied from 8 koe / min to 0.25 koe / min .
the notional positions of onset of smp and pe anomalies have been indicated .
panel ( b ) shows a portion of the two quadrant isothermal m - h loop in sample @xmath27 of @xmath0-nbse@xmath1 at 2 k , measured with a scan rate of 5 koe / min .
an inset in panel ( b ) shows a comparison of the forward legs of the m - h loop in the pe region measured at scan rates of 5 koe / min and 0.5 koe / min , respectively . ] ) data in dc magnetic field of 10 koe , measured with @xmath87 of 2.5 oe ( r.m.s . ) in the frequency range 1 hz to 1111 hz . panel ( b ) shows the @xmath86 data measured at two amplitudes of the ac drive ( @xmath87 of 1.0 oe ( r.m.s . ) and 3.5 oe ( r.m.s . ) ) at a measuring frequency of 111 hz . ] of @xmath0-nbse@xmath1 , for @xmath92 . the re - entrant disordered phase , bragg glass phase , vg phase ( sub - parts i and ii ) and the surface pinning regions have been marked . the spinodal line as in ref .
@xcite has also been drawn . ] | we have investigated the healing of transient disordered vortex states injected into a superconducting sample during the field ramping process in isothermal m - h measurements as a prescription to comprehend the modulation in the state of order of the underlying equilibrium vortex state in weakly pinned single crystals of @xmath0-nbse@xmath1 ( zero field transition temperature , @xmath2 ) and ca@xmath3rh@xmath4sn@xmath5 ( @xmath6 ) , which display order - disorder transition(s ) .
our results reaffirm the notion that spatial order in the flux line lattice proceeds to the amorphous state in two distinct steps as evidenced by the presence of second magnetization peak ( smp ) anomaly and the peak effect ( pe ) phenomenon .
disordering commencing at the onset field of smp has a window to heal between its peak field ( @xmath7 ) and the onset field ( @xmath8 ) of the pe .
a comparison of the scan rate dependence of magnetization hysteresis widths in two different crystals of @xmath0-nbse@xmath1 , one of which displays only the pe phenomenon and the other one that displays both smp and pe anomalies , show that while enhancment in quenched random pinning invokes the occurence of smp anomaly , it slows down the temporal decay of currents across the pe region . |
let @xmath0 with @xmath1 and suppose @xmath2 is a bounded , open set in @xmath3 with locally lipschitz boundary @xmath4 fix @xmath5 such that @xmath6 is bounded and @xmath7 is nondecreasing in @xmath8 for @xmath9 consider the prescribed mean curvature dirichlet problem of finding a function @xmath10 which satisfies @xmath11 where @xmath12 and @xmath13 is a prescribed function ; such a function @xmath14 if it exists , is a classical solution of the dirichlet problem .
it has been long known ( e.g. bernstein in 1912 ) that some type of boundary curvature condition ( which depends on @xmath6 ) must be satisfied in order to guarantee that a classical solution exists for each @xmath13 ( e.g. @xcite ) .
when @xmath15 and @xmath16 is smooth , this curvature condition is that @xmath16 must have nonnegative mean curvature ( with respect to the interior normal direction of @xmath2 ) at each point ( @xcite ) .
however , leon simon ( @xcite ) has shown that if @xmath17 is smooth ( i.e. @xmath18 ) , the mean curvature @xmath19 of @xmath16 is negative on @xmath20 and @xmath21 is a compact subset of @xmath22 then the minimal hypersurface @xmath23 @xmath24 extends to @xmath25 as a continuous function , even though @xmath26 may not equal @xmath27 on @xmath28 since @xcite appeared , the requirement that @xmath15 has been eliminated and the conclusion remains similar to that which simon reached ( see , for example , @xcite ) .
how important is the role of boundary smoothness in the conclusions reached in @xcite ?
we shall show , by constructing suitable domains @xmath2 and dirichlet data @xmath29 that the existence of a `` nonconvex corner '' @xmath30 in @xmath21 can cause the unique generalized ( e.g. variational ) solution to be discontinuous at @xmath30 even if @xmath31 is smooth and the generalized mean curvature @xmath32 ( i.e. @xcite ) of @xmath21 at @xmath30 is @xmath33 ; this shows that some degree of smoothness of @xmath21 is required to obtain the conclusions in @xcite .
we shall prove the following [ main theorem ] let @xmath34 @xmath35 and assume there exists @xmath36 such that @xmath37 for @xmath38 and @xmath39 then there exist a domain @xmath40 and a point @xmath41 such that * @xmath42 is smooth ( @xmath43 ) , * there is a neighborhood @xmath44 of @xmath30 such that @xmath45 for @xmath46 where @xmath19 is the mean curvature of @xmath47 and * @xmath48 where @xmath32 is the generalized mean curvature of @xmath47 and there exists dirichlet boundary data @xmath49 such that the minimizer @xmath50 of @xmath51 exists and satisfies ( [ one - a ] ) , @xmath52 @xmath53 and @xmath54 in a neighborhood of @xmath30 in @xmath4 since there are certainly many examples of dirichlet problems which have continuous solutions even though their domains fail to satisfy appropriate smoothness or boundary curvature conditions ( e.g. by restricting to a smaller domain a classical solution of a dirichlet problem on a larger domain ) , the question of necessary or sufficient conditions for the continuity at @xmath30 of a generalized solution of a particular dirichlet problem is of interest and the examples here suggest ( to us ) that a `` concus - finn '' type condition might yield necessary conditions for the continuity at @xmath30 of solutions ; see [ cfcondition ] .
we view this note as being analogous to other articles ( e.g. @xcite ) which enhance our knowledge of the behavior of solutions of boundary value problems for prescribed mean curvature equations by constructing and analyzing specific examples .
one might also compare theorem [ main theorem ] with the behavior of generalized solutions of ( [ one - a])-([one - b ] ) when @xmath42 is smooth and @xmath55 for @xmath56 ( e.g. @xcite ) and with capillary surfaces ( e.g. @xcite ) .
in this section , we will assume @xmath58 and @xmath59 this allows us to use explicit comparison functions and illustrate our general procedure .
let @xmath2 be a bounded , open set in @xmath60 with locally lipschitz boundary @xmath16 such that a point @xmath30 lies on @xmath16 and there exist distinct rays @xmath61 starting at @xmath30 such that @xmath16 is tangent to @xmath62 at @xmath63 by rotating and translating the domain , we may assume @xmath64 and there exists a @xmath65 such that @xmath66 @xmath67 and @xmath68 for some @xmath69 and functions @xmath70 which satisfy @xmath71 @xmath72 and @xmath73 ; here @xmath74 is the open ball in @xmath60 centered at @xmath30 of radius @xmath75 if we set @xmath76 then @xmath77 and the angle at @xmath30 in @xmath2 of @xmath16 has size @xmath78 as @xmath79 goes to zero , @xmath80 goes to @xmath81 and the ( upper ) region between @xmath82 and @xmath83 becomes `` less nonconvex '' and approaches a half - plane through @xmath63 we will show that for each choice of @xmath84 there is a domain @xmath2 as above and a choice of dirichlet data @xmath85 such that the solution of ( [ one - a])-([one - b ] ) for @xmath2 and @xmath27 is discontinuous at @xmath63 fix @xmath86 let @xmath87 be a small , fixed parameter , say @xmath88 and let @xmath89 be a parameter to be determined
. set @xmath90 and @xmath91 define @xmath92 by @xmath93 notice that the graph of @xmath94 is part of scherk s first surface , so @xmath95 on @xmath96 and @xmath97 for each @xmath98 a computation using lhospital s rule shows @xmath99 let @xmath100 be the intersection of three open disks and let @xmath101 be a strictly convex domain such that @xmath102 is a @xmath43 curve , @xmath103 @xmath104 is symmetric with respect to the @xmath105axis and @xmath106 here @xmath107 denotes @xmath108 define @xmath109 ( see figure [ figureone ] ) ; notice that @xmath110 and ( [ pizza ] ) holds with the choice of @xmath111 above .
if we set @xmath112 then ( [ zipzoom ] ) implies @xmath113 ] let @xmath114 notice that @xmath115 is independent of the parameter @xmath116 define @xmath85 by @xmath117 on @xmath118 and @xmath119 on @xmath120 let @xmath26 be the variational solution of ( [ one - a])-([one - b ] ) with @xmath27 as given here ( e.g. @xcite ) . since @xmath121 on @xmath16 and @xmath122 on @xmath123 @xmath124 in @xmath2 ( e.g. lemma [ four ] ( with @xmath125 ) ) and so @xmath126 in @xmath2 ( e.g. the hopf boundary point lemma ) . notice that @xmath127 on @xmath128 and @xmath129 on @xmath130 and therefore @xmath131 on @xmath132 ( see figure [ figuretwo ] ) .
together with ( [ zipzoom ] ) , this implies @xmath133 the domain of the comparison function for ( [ trex ] ) ] set @xmath134 ( see figure [ figurethree ] ) ; then @xmath135 .
define @xmath136 by @xmath137 and notice that the graph of @xmath138 is part of a catenoid , where @xmath139 on @xmath118 and @xmath140 on @xmath141 it follows from the general comparison principle ( e.g. @xcite , theorem 5.1 ) that @xmath142 on @xmath143 and therefore @xmath144 if we select @xmath145 so that @xmath146 then ( [ trex ] ) and ( [ raptor ] ) imply that @xmath26 can not be continuous at @xmath63 notice that @xcite implies @xmath147 the domain of the comparison function for ( [ raptor ] ) ] this example illustrates the procedure we shall use in [ happy ] ; a somewhat similar approach was used in @xcite .
the case when @xmath148 has a similar proof with the changes that @xmath149 is the intersection of the open disk @xmath150 with the interiors of two ellipses and a scherk surface with rhomboidal domain ( @xcite , pp .
70 - 71 ) is used as a comparison surface to obtain the analog of ( [ trex ] ) ; the details can be found in @xcite .
[ three ] let @xmath2 be a bounded open set in @xmath151 @xmath35 with locally lipschitz boundary and let @xmath21 be an open , @xmath152 subset of @xmath4 let @xmath153 suppose @xmath154 is the variational solution of ( [ one - a])-([one - b ] ) and @xmath155 on @xmath28 then @xmath156 and @xmath157 on @xmath158 where @xmath159 is the exterior unit normal to @xmath21 at @xmath160 * proof : * since @xmath138 minimizes the functional @xmath161 in ( [ the_functional ] ) over @xmath162 @xmath138 also minimizes the functional @xmath163 notice @xmath164 for each @xmath165 with @xmath166 on @xmath167 in particular , this holds when @xmath168 therefore , for each @xmath169 there exists @xmath170 such that @xmath171 and the lemma follows as in @xcite . [ four ] let @xmath2 be a bounded open set in @xmath151 @xmath35 with locally lipschitz boundary , @xmath172 with @xmath173 on @xmath47 @xmath174 with @xmath175 nondecreasing in @xmath8 for @xmath24 and @xmath176 on @xmath177 consider the boundary value problem @xmath178 suppose @xmath154 is the variational solution of ( [ one - a])-([one - b ] ) and either ( i ) @xmath179 is the variational solution of ( [ one - d])-([one - e ] ) or ( ii ) @xmath180 @xmath181 and @xmath182 satisfies ( [ one - d])-([one - e ] )
. then @xmath183 in @xmath184 * proof : * let @xmath185 in case ( i ) , let @xmath186 where @xmath187 is the characteristic function of a set @xmath188 then a simple calculation using @xmath189 shows that @xmath190 and therefore @xmath191 and @xmath192 where @xmath193 @xmath194 is the functional @xmath182 minimizes . case ( ii )
follows from lemma 1 of @xcite .
[ mango ] let @xmath195 be a bounded open set , @xmath0 with @xmath1 and @xmath196 set @xmath197 and define @xmath198 by @xmath199 for @xmath200 then , for @xmath201 with @xmath202 @xmath203 and @xmath204 we have @xmath205 in particular , if @xmath206 @xmath207 @xmath208 and @xmath209 then @xmath210 on @xmath211 * proof : * notice that @xmath212 @xmath213 @xmath214 and so @xmath215 @xmath216 the lemma follows from this .
let @xmath218 denote the open ball in @xmath219 centered at @xmath220 with radius @xmath221 and @xmath222 for @xmath223 now consider @xmath1 and set @xmath224 if @xmath225 replace it with a positive constant .
for each @xmath226 and @xmath227 we have @xmath228 by translating our problem in @xmath151 we may ( and will ) assume @xmath229 by proposition 1.1 and theorem 2.1 of @xcite , we see that if @xmath2 is a bounded , connected , open set in @xmath3 with lipschitz - continuous boundary , @xmath230 and @xmath231 then the functional @xmath161 in ( [ the_functional ] ) has a minimizer @xmath232 @xmath233 and @xmath26 satisfies ( [ one - a ] ) .
the proof in [ cdo ] consists of setting some parameters ( e.g. @xmath234 @xmath235 @xmath236 @xmath237 @xmath238 @xmath239 @xmath240 @xmath241 @xmath242 ) , determining the domain @xmath243 finding different comparison functions ( e.g. @xmath244 @xmath245},$ ] @xmath246
@xmath247 @xmath248 @xmath249 ) , and mimicking ( [ trex ] ) and ( [ raptor ] ) to show that the variational solution @xmath26 of ( [ one - a])-([one - b ] ) is discontinuous at a nonconvex corner . in particular , we use a torus ( i.e. @xmath250 ) to obtain ( [ hopping ] ) , unduloids ( i.e. @xmath246 @xmath251 ) to obtain ( [ catx ] ) ( an analog of ( [ raptor ] ) ) and nodoids ( i.e. @xmath244 @xmath245}$ ] ) , unduloids ( i.e. @xmath246 @xmath249 ) and a helicoidal function ( i.e. @xmath252 ) to obtain ( [ cab ] ) ( an analog of ( [ trex ] ) ) and prove that @xmath26 is discontinuous at @xmath253 in this section , we will obtain a domain @xmath2 as above and @xmath255 such that @xmath256 the minimizer @xmath26 of ( [ the_functional ] ) is discontinuous at @xmath257 @xmath258 is smooth ( @xmath43 ) and @xmath259 where @xmath260 is a smooth set of dimension @xmath261 ( i.e. @xmath260 has codimension @xmath254 in @xmath16 ) .
we will use portions of nodoids , unduloids and helicoidal surfaces with constant mean curvature as comparison functions . for the convenience of the reader
, we will denote functions whose graphs are subsets of nodoids with the letter @xmath138 ( e.g. @xmath262 ) , subsets of cmc helicoids with the letter @xmath182 and subsets of unduloids ( or onduloids ) with the letter @xmath263 let @xmath264 be a nodoid which is symmetric with respect the @xmath265-axis and has mean curvature @xmath254 ( when @xmath266 is oriented `` inward '' , so that the unit normal @xmath267 to @xmath266 points toward the @xmath265-axis at the points of @xmath266 which are furthest from the @xmath265-axis ) .
let @xmath268 be the inner neck size of @xmath266 and let @xmath269 satisfy the condition that the unit normal to @xmath266 is vertical ( i.e. parallel to the @xmath265-axis ) at each point @xmath270 of @xmath266 at which @xmath271 then @xmath272 let @xmath273 ( notice that we can assume @xmath274 is close to @xmath275 if we wish . )
let us fix @xmath276 and set @xmath277 @xmath278 let @xmath279 we shall assume @xmath280 if necessary , we increase @xmath281 to accomplish this .
let @xmath282 and @xmath283 let @xmath284 then @xmath44 is a nodoid with mean curvature @xmath285 set @xmath286 fix @xmath287 define @xmath288 to be a function whose graph is a subset of @xmath44 on which @xmath289 satisfies @xmath290 then @xmath291 by moving @xmath44 vertically , we may assume @xmath292 when @xmath293 then @xmath294 in @xmath295 notice that @xmath296 and @xmath297 ; then there exists a @xmath298 such that , for each @xmath299 @xmath300 where @xmath301 fix @xmath302 let @xmath303 consider @xmath304 notice that the distance between @xmath305 and the point @xmath306 is @xmath307 where @xmath305 is the closed sector given by @xmath308 define @xmath309 and @xmath310 notice that @xmath311 and therefore @xmath312 if @xmath313 set @xmath314 then @xmath315 since @xmath316 and @xmath317 where @xmath318 therefore there exists a @xmath319 such that if @xmath320 with @xmath321 then @xmath322 since @xmath323 we have @xmath324 and so @xmath325 ( see figure [ figureeight ] ( b ) ) . notice that @xmath326 for some functions @xmath70 which satisfy @xmath71 @xmath327 and @xmath328 let @xmath329 and set @xmath330 where @xmath331\times [ -\pi,0]$ ] and @xmath332 since @xmath333 we see that @xmath334 for all @xmath335 we shall assume @xmath336 and @xmath337 notice that @xmath338 is the lower half of a torus whose mean curvature ( i.e. one half of the trace of the shape operator ) at each point is greater than @xmath285 let @xmath338 be the graph of a function @xmath250 over @xmath339 then @xmath340 on @xmath341 and @xmath342 @xmath343 on @xmath344 and @xmath345 on @xmath346 and @xmath347 for @xmath348 notice that @xmath349 for all @xmath350 set @xmath351 where @xmath352 if we define @xmath353 for @xmath354 and @xmath355 then @xmath356 for @xmath357 and @xmath358 for @xmath359 ( see figure [ figurefive ] ) . for @xmath357 ( b ) @xmath360 for @xmath359 ]
we wish to select a helicoidal surface in @xmath57 ( e.g. @xcite ) with constant mean curvature @xmath237 axis @xmath362 and pitch @xmath363 ( recall @xmath364 , which we will denote @xmath365 then , for each @xmath366 @xmath367 where @xmath368 is the helicoidal motion given by @xmath369 with @xmath370 given by @xmath371 set @xmath372 by vertically translating @xmath373 we may assume that there is an open @xmath374level curve @xmath375 of @xmath376 with endpoints @xmath377 and @xmath378 such that @xmath379 @xmath380 is tangent to the ( horizontal ) line @xmath381 at @xmath382 and the slope @xmath383 of the tangent line to @xmath384 at @xmath385 satisfies @xmath386 for each @xmath387 then @xmath388 and the curves @xmath389 @xmath390 are mutually disjoint .
notice that the set @xmath391 is an open subset of @xmath392\times \{p\}\right)$ ] ( see figure [ figuresix ] ) , @xmath393 and @xmath376 implicitly defines the smooth function @xmath252 on @xmath394 given by @xmath395 if @xmath396 for some @xmath397 notice
that @xmath398 now @xmath399 for @xmath400 and , by making @xmath401 sufficiently small , we may assume that @xmath402 notice that @xmath403 on @xmath404 for @xmath405 ] let us fix @xmath320 such that @xmath406 and @xmath407 then there exists @xmath408 such that @xmath409 define @xmath245}(x)=g_{1}\left(x+\frac{r_{2}-p}{p}u\right)$ ] and notice that @xmath245}(u)=g_{1}\left(\frac{r_{2}}{p}u\right)=0,$ ] since @xmath410 note that the domain @xmath411}= \{x+\frac{p - r_{2}}{p}u : x\in \delta_{1}\ } = b_{2}\left(\frac{p - r_{2}}{p}u , r_{2}\right ) \setminus \overline { b_{2}\left(\frac{p - r_{2}}{p}u , r_{1}\right ) } \ ] ] of @xmath245}$ ] is contained in @xmath412 since @xmath413 and @xmath414 are tangent circles at @xmath415 and @xmath416 ( see figure [ figureseven ] ) .
notice that @xmath417}(r\cos(\theta_{u}),r\sin(\theta_{u}))\ ] ] when @xmath418 because @xmath419}(u),$ ] @xmath420 and ( [ dog ] ) holds .
let @xmath421 be unduloids in @xmath57 with mean curvature @xmath422 such that @xmath423 are the respective axes of symmetry ; the minimum and maximum radii ( or `` neck '' and `` waist '' sizes ) of both unduloids are @xmath424 and @xmath425 respectively . set @xmath426 and define @xmath427 so that the graphs of @xmath428 are subsets of @xmath429 respectively , @xmath430 @xmath431 and @xmath432 for each @xmath433 where @xmath301 we may vertically translate @xmath429 so that @xmath434 for @xmath435 with @xmath436 notice that @xmath437 ( b ) @xmath438 ] let @xmath439 be an unduloid with mean curvature @xmath422 such that the @xmath440axis is the axis of symmetry and the minimum and maximum radii ( or `` neck '' and `` waist '' sizes ) are @xmath441 and @xmath442 respectively . set @xmath443 and define @xmath444 so that the graph of @xmath251 is a subset of @xmath445 @xmath446 in @xmath447 @xmath448 and @xmath449 for each @xmath433 where @xmath301 ( [ once upon a time ] ) ] define @xmath49 so that @xmath117 on @xmath450 and @xmath119 on @xmath451 where @xmath452 recall then that @xmath453 let @xmath26 be the variational solution of ( [ one - a])-([one - b ] ) with @xmath2 and @xmath27 as given here ; that is , let @xmath26 minimize the functional given in ( [ the_functional ] ) and notice that the existence of @xmath26 follows from ( [ book ] ) , ( [ assumption_a ] ) ,
1.d . of @xcite and @xcite .
( notice that there exists @xmath454 such that @xmath455 ) the comparison principle implies @xmath456 for @xmath457 and so @xmath458 if @xmath457 with @xmath459 ( recall ( [ assumption_a ] ) holds ) .
in particular , @xmath460 set @xmath461 now @xmath462 ( see figure [ figurenine ] ) .
define @xmath463 for @xmath464 notice that @xmath465 on @xmath466 @xmath467 and @xmath448 ( so that @xmath468 for @xmath469 where @xmath470 is the unit exterior normal to @xmath471 ) .
the general comparison principle ( e.g. @xcite , theorem 5.1 ) then implies @xmath472 and , in particular , @xmath473 ( see figure [ figureten ] ) . by rotating the axis of symmetry of @xmath143 through all lines in @xmath3 containing @xmath474 ( or , equivalently , keeping @xmath143 fixed and rotating @xmath2 about @xmath474 )
, we see that @xmath475 ): @xmath143 and @xmath476 when @xmath477 now define @xmath478 by @xmath479 combining ( [ one - a ] ) and ( [ catx ] ) with the facts that @xmath480 in @xmath481 and @xmath482 for @xmath483 where @xmath484 is the inward unit normal to @xmath485 we see that @xmath486 ( if figure [ figureeight ] ( a ) held , then ( [ caty ] ) would not be valid . )
now let @xmath487 be any rotation about @xmath474 which satisfies @xmath488 notice that @xmath489 satisfies ( [ one - a])-([one - b ] ) and apply the previous argument to obtain @xmath490 in @xmath491 and therefore @xmath492 ( [ jerry ] ) ] from lemma [ three ] ,
we see that the downward unit normal to the graph of @xmath14 @xmath493 satisfies @xmath494 on @xmath495 and @xmath496 let us write @xmath497 then @xmath498}=0 \le f$ ] on @xmath499 and @xmath498}\le g_{1}(r_{1},0)<\phi$ ] on @xmath500 it follows from ( [ one - a ] ) , ( [ star ] ) and lemma [ mango ] that @xmath501}<f \ \ \ \ \ { \rm on } \ \ \ \ \ \omega\cap \tilde { \cal d}^{[u]}=\omega \cap \tilde b.\ ] ] set @xmath502 if we write @xmath503,\omega\in s^{n-2 } \ } , $ ] @xmath504 and @xmath505,\omega\in s^{n-2 } \ } , $ ] then @xmath506 @xmath507 on @xmath508 and @xmath509}<f$ ] on @xmath510 ( see ( [ spaceship ] ) ) ; then ( [ doggy ] ) and the general comparison principle imply @xmath511 where @xmath512 ( see figure [ figureeleven ] ) .
set @xmath513 now ( [ starshine ] ) implies @xmath514 and so @xmath515 using ( [ catx ] ) and ( [ cab ] ) , we see that if @xmath516 is close enough to @xmath234 then @xmath517 and therefore @xmath26 can not be continuous at @xmath30 or at any point of @xmath518 .
notice that @xmath519 ( e.g. @xcite ) .
( blue region ) and @xmath520 ( green and blue regions ) ] in this section , we will obtain a domain @xmath2 and @xmath255 such that @xmath256 the minimizer @xmath26 of ( [ the_functional ] ) is discontinuous at @xmath257 @xmath521 is smooth ( @xmath43 ) and @xmath147 this is accomplished by replacing @xmath522 by a convex set @xmath523 such that @xmath524 is smooth ( @xmath43 ) and @xmath525 we shall use the notation of [ cdo ] throughout this section .
we assume @xmath526 and set @xmath527 ( we will no longer require figure [ figureeight ] ( b ) to hold . )
let @xmath528 @xmath529 and @xmath530 \times [ 0,\pi ] \times s^{n-3}\to \real^{n}$ ] be defined by @xmath531 let @xmath532 be given by @xmath533 and define @xmath534 for @xmath535 ( see figures [ figuretwelve ] and figure [ figurethirteen ] with @xmath536 @xmath537 the axes are labeled @xmath538 for @xmath539 respectively ) .
let @xmath523 be the open , convex set whose boundary is the image of @xmath540 that is , @xmath541 notice that @xmath524 is a @xmath43 hypersurface in @xmath3 and @xmath542 , title="fig:",width=151 ] , title="fig:",width=151 ] let @xmath543 satisfy @xmath544 set @xmath545 and @xmath546 then the tangent cones to @xmath547 and @xmath548 at @xmath30 are identical , @xmath549 and ( [ cat2 ] ) holds for @xmath320 with @xmath550 if necessary by making @xmath551 smaller , we may assume @xmath552 if @xmath313 now pick @xmath553 such that @xmath554 as in ( [ cab ] ) , and define @xmath555 let @xmath556 and define @xmath49 so that @xmath117 on @xmath450 and @xmath119 on @xmath547 and let @xmath26 be the variational solution of ( [ one - a])-([one - b ] ) . notice that @xmath233 satisfies ( [ one - a ] ) and @xmath557 ( e.g. @xcite ) . as in ( [ bingo ] )
, let @xmath558 set @xmath559 and @xmath560 now @xmath498}=0 $ ] on @xmath561 and @xmath498}\le g_{1}(0,r_{1})<m$ ] on @xmath562 and so lemma [ four ] , lemma [ mango ] and ( [ one - a ] ) imply @xmath498 } \le f$ ] in @xmath563 since @xmath26 minimizes the functional in ( [ the_functional ] ) .
as before , set @xmath564,\omega\in s^{n-2 } \},$ ] @xmath565 and @xmath505,\omega\in s^{n-2 } \}.$ ] then @xmath124 on @xmath566 @xmath506 @xmath507 on @xmath567 @xmath568 on @xmath569 and @xmath509}<f$ ] on @xmath570 lemma [ four ] implies that ( [ cab ] ) continues to hold
. then ( [ catx ] ) and ( [ cab ] ) imply @xmath26 is discontinuous at @xmath30 since @xmath571 ( b ) @xmath572 , title="fig:",width=188 ] ( b ) @xmath572 , title="fig:",width=188 ]
for the moment , assume @xmath573 in approximately 1970 , paul concus and robert finn conjectured that if @xmath574 @xmath575 has a corner at @xmath41 of ( angular ) size @xmath576 @xmath577 @xmath578 $ ] and @xmath579 where @xmath580 then a function @xmath581 which satisfies @xmath582 must be discontinuous at @xmath583 here @xmath584 is the exterior unit normal to @xmath2 at @xmath585 a generalization ( including the replacement of ( [ ten - a ] ) by ( [ one - a ] ) ) of this conjecture in the case @xmath586 was proven in @xcite .
in the situation above with @xmath587 the `` nonconvex concus - finn conjecture '' states that if @xmath588 then the capillary surface @xmath26 with contact angle @xmath589 must be discontinuous at @xmath63 a generalization ( including the replacement of ( [ ten - a ] ) by ( [ one - a ] ) ) of this extension of the concus - finn conjecture in the case @xmath586 was proven in @xcite .
both @xcite and @xcite include the possibility of differing limiting contact angles ; that is , the following limits @xmath590 exist , @xmath591 and @xmath592 here @xmath593 and @xmath594 are the two components of @xmath595 where @xmath596 when @xmath597 the necessary and sufficient ( when @xmath598 ) or necessary ( when @xmath599 ) conditions for the continuity of @xmath26 at @xmath30 become slightly more complicated . the cases where @xmath600 @xmath601 @xmath602 and @xmath603 remain unresolved . if we suppose for a moment that the nonconvex concus - finn conjecture with limiting contact angles of zero or @xmath81 is proven , then the discontinuity of @xmath26 at @xmath30 in [ blue ] follows immediately from the fact that @xmath604 in a neighborhood in @xmath521 of @xmath30 since then lemma [ three ] implies @xmath605 and therefore @xmath606 in this situation ( i.e. the solution @xmath26 of a dirichlet problem satisfies a zero ( or @xmath81 ) contact angle boundary condition near @xmath30 ) , establishing the discontinuity of @xmath26 at @xmath30 would be much easier and a much larger class of domains @xmath2 with a nonconvex corner ( i.e. @xmath599 ) at @xmath30 would have this property .
for example , if @xmath2 is a bounded locally lipschitz domain in @xmath60 for which ( [ pizza ] ) holds , @xmath233 is a generalized solution of ( [ one - a])-([one - b ] ) ( and @xmath6 need not vanish ) and @xmath27 is large enough near @xmath30 ( depending on @xmath6 and the maximum of @xmath27 outside some neighborhood of @xmath30 ) that @xmath604 on @xmath521 near @xmath257 then the fact that @xmath605 ( lemma [ three ] ) together with the nonconvex concus - finn conjecture would imply that @xmath26 is discontinuous at @xmath63 now consider @xmath607 with @xmath608 formulating generalizations of the concus - finn conjecture in the `` convex corner case '' ( i.e. @xmath609 for some @xmath610 @xmath41 and @xmath221 ) and in other cases where @xmath16 is not smooth at a point @xmath41 may be complicated because the geometry of @xmath521 is much more interesting when @xmath611 establishing the validity of a generalization of the concus - finn conjecture for solutions of ( [ one - a ] ) & ( [ fletch ] ) when @xmath612 is probably significantly harder than doing so when @xmath573 suppose we knew that a solution @xmath26 of ( [ one - a ] ) & ( [ fletch ] ) is necessarily discontinuous at a `` nonconvex corner '' @xmath41 when @xmath600 where @xmath613 is given by ( [ frozen ] ) . in this case , a necessary condition for the continuity of @xmath26 at @xmath30 would be that @xmath614 and @xmath615 then the arguments in [ happy ] could be made more easily and the conclusion that @xmath26 is discontinuous at @xmath30 would hold in a much larger class of domains @xmath616 here , of course , we use the ridge point @xmath30 in [ happy ] as an example of a `` nonconvex corner '' of a domain in @xmath617 the primary difficulty in proving in [ happy ] that @xmath26 is discontinuous at @xmath30 is establishing ( [ cab ] ) ; a more `` natural '' generalization of @xmath618 in [ blue ] would be @xmath619 however , the use of lemma [ mango ] to help establish ( [ cab ] ) in @xmath620 is highly problematic .
on the other hand , an n - dimensional `` concus - finn theorem '' for a nonconvex conical point ( e.g. @xmath621 ) would only require an inequality like ( [ taxi ] ) to prove that @xmath604 on @xmath42 near @xmath30 and hence that @xmath26 is discontinuous at @xmath583 the replacement of ( [ aaaa ] ) by ( [ bbbb ] ) in order to obtain a @xmath2 such that @xmath42 is @xmath43 would be unnecessary .
99 t. bourni , _ @xmath622 theory for the prescribed mean curvature equation with dirichlet data _ , journal of geometrical analysis * 21 * ( 2011 ) , 9821035 .
m. do carmo and m. dajczer , _ helicoidal surfaces with constant mean curvature _ , tohoku mathematical journal * 34 * ( 1982 ) , no .
3 , 425435 . a. elcrat and k. lancaster , _ boundary behavior of a nonparametric surface of prescribed mean curvature near a reentrant corner _ ,
* 297 * ( 1986 ) , no .
2 , 645650 .
robert finn , _ equilibrium capillary surfaces _ , vol 284 of grundlehren der mathematischen wissenschaften , springer - verlag , new york , 1986 .
robert finn and d. shi , _ on a theorem of lancaster and siegel _ , pacific journal of mathematics * 213 * ( 2004 ) , no .
1 , 111119 . c. gerhardt , _
existence , regularity , and boundary behavior of generalized surfaces of prescribed mean curvature _ , mathematische zeitschrift * 139 * ( 1974 ) , issue 2 , 173198 .
e. giusti , _ boundary value problems for non parametric surfaces of prescribed mean curvature _ , annalidella seuola normale superiore di pisa * 3 * ( 1976 ) , issue 3 , 501548 .
e. giusti , _ on the equation of surfaces of prescribed mean curvature . existence and uniqueness without boundary conditions _ ,
* 46 * ( 1978 ) , no .
2 , 111137 .
r. huff and j. mccuan , _ scherk - type capillary graphs _ , j. math . fluid mech . * 8 * ( 2006 ) , no
. 1 , 99119 .
r. huff and j. mccuan , _ minimal graphs with discontinuous boundary values _
, j. aust .
* 86 * ( 2009 ) , no .
1 , 7595 .
h. jenkins and j. serrin , _ the dirichlet problem for the minimal surface equation in higher dimensions _
, j. reine angew . math .
* 229 * ( 1968 ) , 170187 .
n. korevaar , _ on the behavior of a capillary surface at a re - entrant corner _ , pacific journal of mathematics * 88 * ( 1980 ) , no .
2 , 379385
. n. korevaar and l. simon , _ equations of mean curvature type with contact angle boundary conditions _ , geometric analysis and the calculus of variations , 175201 , int . press , cambridge , ma , 1996 .
k. lancaster , _ boundary behavior of a nonparametric minimal surface in @xmath57 at a nonconvex point _ ,
analysis * 5 * ( 1985 ) , no . 12 , 6169 .
k. lancaster , _ nonparametric minimal surfaces in @xmath57 whose boundaries have a jump discontinuity _ , internat .
j. math . math .
* 11 * ( 1988 ) , no .
4 , 651656 .
k. lancaster , _ a proof of concus - finn conjecture _ , pacific journal of mathematics , 2010 , * 247 * , no.1 , 75108 .
k. lancaster , _
remarks on the behavior of nonparametric capillary surfaces at corners _ , pacific journal of mathematics , 2012 , * 258 * , no.2 , 369392
. k. lancaster and d. siegel , _ existence and behavior of the radial limits of a bounded capillary surface at a corner _ , pacific journal of mathematics , 1996 , * 176 * , no .
1 , 165194 . c p lau and f h lin , _ the best hlder exponent for solutions of the non - parametric least area problem _ , indiana u. math .
journal , * 34 * ( 1985 ) , no .
4 , 809813 .
f h lin , _ behaviour of nonparametric solutions and free boundary regularity _ , miniconference on geometry and partial differential equations , 2 , canberra , 1986 , proceedings of the centre for mathematical analysis , australian national university * 12 * , pp .
australian national university , canberra ( 1987 ) .
j. melin , _ examples of discontinuity for the variational solution of the minimal surface equation with dirichlet data on a domain with a nonconvex corner and locally negative mean curvature _ , ms thesis , wichita state university ( 2013 ) .
nitsche , _ lectures on minimal surfaces _ , vol . 1 , cambridge university press , cambridge , 1989 .
j. serrin , _ the problem of dirichlet for quasilinear elliptic differential equations with many independent variables _ , phil .
lond . a * 264 * ( 1969 ) , no . 1153 ,
. l. simon , _ boundary regularity for solutions of the non - parametric least area problem _ , annals of math .
* 103 * ( 1976 ) , 429455 .
g. williams , _ surfaces of prescribed mean curvature with inequalities on the boundary _ , math . z.
* 164 * ( 1978 ) , no .
1 , 3151 . | in 1976 , leon simon showed that if a compact subset of the boundary of a domain is smooth and has negative mean curvature , then the non - parametric least area problem with lipschitz continuous dirichlet boundary data has a generalized solution which is continuous on the union of the domain and this compact subset of the boundary , even if the generalized solution does not take on the prescribed boundary data .
simon s result has been extended to boundary value problems for prescribed mean curvature equations by other authors . in this note , we construct dirichlet problems in domains with corners and demonstrate that the variational solutions of these dirichlet problems are discontinuous at the corner , showing that simon s assumption of regularity of the boundary of the domain is essential . keywords : nonconvex corner , prescribed mean curvature
. 2010 msc : primary 35j93 , 76b45 ( secondary 35j62 , 53a10 ) |
globular cluster systems rank among the oldest objects known in our galaxy .
their ages set a lower limit for the age of the universe .
unfortunately , a precise measurement of the absolute age of a globular cluster ( gc ) is strongly dependent on the knowledge of its precise distance and reddening ( @xcite ; @xcite ) .
m4 is a peculiar cluster in terms of interstellar reddening .
it has a surprisingly large total amount of reddening ( @xmath13 , @xcite ) for its small distance ( @xmath14 ) , which is due to its location in the galactic plane , behind the sco - oph cloud complex .
moreover , the cluster suffers from a significant amount of spatially differential reddening ( @xcite ; @xcite ; and @xcite ) which has been taken into account in some recent studies of m4 ( e.g. @xcite , @xcite ) . in the literature ,
approximations for peak - to - peak differences within a distance to the cluster centre of at least @xmath15 range from @xmath16 in @xcite to @xmath17 in @xcite .
the extinction law that is used to correct for interstellar reddening can be different for each line of sight , depending on the type of dust located between the observer and the object , where the total to selective extinction @xmath18 is known to be a good index of the extinction properties of a particular dust type ( @xcite , @xcite ) .
however , in many studies of star clusters a standard reddening law is applied , assuming dust characteristics typical of the diffuse interstellar medium ( ism ) as the main origin of interstellar extinction : a reddening law with a value of @xmath0 close to 3.1 is typically adopted ( e.g. @xcite , @xcite ) . however , since @xmath0 defines both the shape of the reddening law and the ratio between extinction and reddening , a reconsideration of this standard assumption is appropriate when determining the absolute distance of a reddened stellar system . for m4
, there are several hints in former studies where the authors suggest an abnormal dust type , or at least remark on certain discrepancies that arise when using a standard assumption of @xmath19 .
for example , @xcite compare their spectroscopically derived temperatures of bright red giant branch ( rgb ) stars to the temperatures derived from photometric indices and found @xmath20 .
@xcite propose an @xmath0 closer to 4 then to 3 " by evaluating the relative location of rgb sequences from m4 compared to m3 , m13 and m92 using data from @xcite .
interestingly , the only study that does not rely on the calibration of photometric magnitudes reaches a similar result : @xcite derived a geometric distance for m4 from proper motion and radial velocity measurements and find a value significantly smaller than the accepted photometric distance , a result which is only in agreement with canonical horizontal branch ( hb ) magnitude measurements if a reddening law with @xmath21 is assumed .
in addition to direct measurements of @xmath0 using cluster members , there are several indirect measurements using individual stars in the sco - oph dust cloud complex with a line of sight close to m4 , which yield values of @xmath0 around 4 ( see e.g. , @xcite , @xcite ) .
indeed , some studies of m4 s colour - magnitude diagram ( cmd ) already use a non - standard , higher value of @xmath0 .
@xcite ( and later @xcite ) were among the first to use a value of @xmath22 for their distance estimate when determining the age of m4 from the white dwarf cooling sequence .
however , this choice is solely based on the vague statements of @xcite , and @xcite who only estimate the value to be around 4 . " later , @xcite and @xcite follow the argumentation of richer and additionally use the results from @xcite , who found a value of @xmath22 for a star only one degree away from the line of sight of m4 to justify their choice
. however , @xcite warns in his review about the properties of interstellar dust that it is not possible to estimate @xmath0 quantitatively from the environment of a line of sight , " since significant variations in the properties of dust can occur even on these small angular scales ( see also @xcite ) .
moreover , this value for @xmath0 is based on the measurement of only one star , using obsolete stellar models to estimate its atmospheric parameters .
m4 is the closest gc to the sun .
its closeness and low degree of crowding make the stellar population accessible to especially deep photometric and spectroscopic analyses .
for example , m4 hosts one of the largest populations of identified white dwarfs ( wds ) , making it attractive for absolute age determinations and an excellent laboratory for testing stellar evolutionary theory ( e.g. @xcite ) .
it has a significant population of rr lyrae stars and it shows a bimodal hb , with well populated blue and red parts , making it an interesting object to study the so called second parameter problem " ( see e.g. the review by @xcite or some recent work by @xcite ) .
there is clearly a need for a detailed investigation of the absolute reddening and the specific reddening law for m4 , providing more precise information about its dust type and the effects of spatial differential reddening across the cluster face . a better determination of the absolute photometric properties of m4 s stars and consequently its distance will increase the cluster s status as a most attractive target for astrophysical investigations .
in this study we investigate the type of dust in the line of sight to m4 , as characterized by the appropriate value of @xmath0 assuming the validity of the general reddening law given in @xcite .
we achieve high quantitative precision by using a combination of near infrared ( nir ) @xmath23 and @xmath24 and optical johnson - cousins @xmath25 photometry for the cluster m4 , supplemented with equivalent photometry for the cluster ngc 6723 with which we verify our models in unreddened conditions .
we use the newest set of victoria - regina isochrones ( @xcite ) to determine colour excesses in filter combinations that are sensitive to the type of dust and consequently to the shape of the reddening law . to increase the precision of parameters such as distance , age and the absolute locus of the observed fiducial sequence , we map the variations in the reddening across the field of m4 that has been surveyed and then correct for those differential effects . since extinction is a function of wavelength , the accurate extinction properties of our broadband filters depend upon the distribution of observed stellar flux within the passband . to minimize systematic errors in our extinction corrections and to the determined value of @xmath0
, we investigate differential changes in the extinction properties of optical and nir filters and filter combinations appropriate for stars with different intrinsic spectral - energy distributions as determined by their temperature , surface gravity and metallicity .
specifically , we consider whether a single reddening law is sufficient to determine cluster parameters from photometry , or whether a star - by - star correction is required to avoid significant systematic errors .
these results are then used to define an object - specific reddening law for m4 where the correction zero points are tailored for the atmospheric and chemical parameters of this cluster instead of assuming a vega - like star , as is the case for most literature values . to test the consistency of the results obtained with synthetic and observed stellar fluxes we compare synthetic spectra from the atlas9 library ( @xcite ) to observational databases from @xcite and @xcite .
the outcome of such a comparison not only constrains the validity of our results , but also reveals possible weak spots in the atmospheric models . in [ chapter_datareduction ] ,
we summarize the data and the reduction process together with the criteria for choosing our high - quality photometric sample . our procedure for generating a reddening map for m4 and the spatial differential reddening corrections is explained in [ chapter_diffred ] , while in [ chapter_reddeninglaw ] , we investigate the effects of stellar temperature , surface gravity , metallicity and extinction on the shape of the reddening law and calculate an object - specific law tailored for m4 . finally , in [ chapter_dusttype ] we characterize the dust type toward m4 , derive the cluster s absolute distance , and discuss random and systematic errors in detail .
a summary is given in
[ chapter_summary ] .
appendices provide fiducial sequences for m4 and ngc 6723 , and explicit extinction tables for a wide range of @xmath26 , [ fe / h ] and @xmath0 .
for this work , we use near infrared @xmath23 and @xmath24 and @xmath24 are used interchangeably where both refer to the 2mass bandpass . ] ground - based photometry for the globular clusters m4 ( ngc 6121 ) and ngc 6723 .
the images for m4 were taken in 2002 with the large and small field of sofi , the infrared imaging camera on the new technology telescope of eso on cerro la silla in chile as part of the program 69.d-0604(a ) ( pi : zoccali ) and cover a field of @xmath27 . for ngc 6723
the same telescope and instrumentation were used in 2005 as part of the program 075.d-0372 ( pi : ferraro ) .
the pre - processing of the ccd images for both clusters included bad pixel masking , bias and dark frame subtraction , and flat - field corrections , as well as sky subtraction .
we use the data reduction packages daophot ii ( @xcite ; @xcite ) and allframe ( @xcite ) and related routines to obtain the instrumental photometry for each cluster and transform it to the 2mass photometric system ( @xcite ) . since the data reduction process for our two clusters is very similar , we will describe this work in detail only for m4 as an example . in total , 50 frames in each filter
were taken with a main concentration on the central region of the cluster , which was observed 10 times , and one region outside the core , which was observed 20 times .
the rest of the cluster is covered by only one exposure in each filter with slight overlap . for the profile - fitting photometry
we use a point spread function ( psf ) which varies quadratically with the position in the frame because the sofi large - field camera was not properly aligned , so that the images are radially elongated in the left part of each frame .
this distortion affects a vertical strip of about 150 pixels and smoothly disappears toward the centre of the frame .
later , we correct the residual spatial inhomogeneities to first order by applying a differential reddening correction ( see section [ chapter_diffred ] ) . the psf is calculated using a sample of bright , isolated program stars , uniformly distributed over the individual frames . although we have written a pipeline to process all frames through daophot and allframe automatically , we pick out the psf stars for each frame by eye to assure their quality and rule out blending .
the same stars later serve as local standards to calibrate our instrumental photometry to the 2mass standard system .
we use daomatch and daomaster to find positional transformation equations between the frames , accounting for differences in scale and rotation as well as transverse offsets .
those equations are then used to match up stars between different frames in order to get a complete star list for our catalog . to calibrate the frames among themselves , daogrow ( @xcite )
is used to obtain the total integrated instrumental magnitudes ( i.e. , the total photon count ) for our standard stars by multiple aperture photometry .
these total magnitudes are then calibrated by direct comparison to the published 2mass magnitudes for the same stars , as described in the next section .
our instrumental photometry for m4 must be calibrated to the 2mass standard system : for accurate comparisons of our observed data to stellar evolutionary models , it is especially important that systematic deviations between instrumental and standard magnitudes are small .
the same hand - selected , bright and isolated program stars used to determine the psf above now serve as local standards for our photometry . as an additional criterion , we only use stars below the 1% linearity limit of the detector ( @xmath28 ) , and only those that could be cross - identified with a 2mass standard star with a coordinate deviation less than @xmath29 . for m4 ,
these comprise about 360 stars covering the whole observation field . since they are distributed homogeneously over the different frames , we are able to detect and compensate for photometric differences between different exposures as well as spatial inhomogeneities within individual fields . and @xmath30 .
only stars are shown that could be cross - identified with a 2mass star to within @xmath29 .
the solid black line indicates a theoretical zero offset .
stars used as local standards are viewed in black , grey dots indicate all other identified 2mass stars from our final high - quality sample ( see following paragraphs).,scaledwidth=50.0% ] to examine whether there are any trends with temperature.,scaledwidth=50.0% ] daomaster is then used to match up 2mass stars with corresponding local standards in our photometry .
the equation that we use to transform the instrumental magnitudes ( @xmath31 ) to the standard system ( @xmath32 ) is of the general form @xmath33 and includes a zero - point correction ( @xmath34 ) , a term for atmospheric extinction ( @xmath35 ; with airmass @xmath36 ) , and a colour - sensitive term ( @xmath37 ) to account for differences in filter characteristics between the two systems . for our @xmath23 photometry
we find @xmath38 , @xmath39 , and @xmath40 . for the @xmath41 filter , @xmath42 , @xmath43 , and @xmath44 .
the correction coefficients for atmospheric extinction ( @xmath35 ) are a little low for @xmath23 and quiet high for @xmath41 , but since the variation in airmass within our individual frames is only 0.028 for both nir bandpasses , the impact on our photometry compared to `` standard '' assumptions of @xmath45 for each filter would only be in the order of @xmath46 mag and therefore negligible for our results .
once the transformation equations to the standard system are defined for each filter with the standard star sample , they are applied to all of the stars in the cluster . a comparison between our photometry and the 2mass standard system is shown in figures [ n6121_consistency1 ] and [ n6121_consistency2 ] where the deviation between the two systems is plotted as a function of magnitude and colour .
we find neither a significant zero - point offset between the two systems nor a clear trend with colour or magnitude , which would arise from insufficient correction of the different filter characteristics .
the optical johnson - cousins photometry for m4 is a compilation of 4794 , primarily archival , images from 48 independent nights distributed among 11 different observing runs .
the images were obtained between 1994 and 2007 with several different ground - based telescopes ranging from 0.9 m to 3.6 m , and were analyzed and calibrated as described in @xcite and @xcite . in total , the optical database covers a field of @xmath47 around the cluster centre , which is about four times the area that we cover with our nir photometry .
the database consists of @xmath48 stars and the photometry reaches down to @xmath49 , @xmath50 , and @xmath51 .
we use daomaster to match up stars between the two photometric sets , where only stars are considered for which the coordinate deviation between the optical and the nir database is smaller than @xmath29 .
the initial output of the photometry packages daophot and allframe yields a total of @xmath52 stars with photometric information in both @xmath23 and @xmath24 and in at least two of the optical bands .
this sample still includes stars with high photometric errors , field stars in the line of sight to the actual gc system , and variable stars .
since we are more interested in photometric precision than in completeness , a high - quality subsample ( hereafter : hq - sample ) was selected taking into consideration the aspects mentioned above .
to select only stars with the best photometric standard errors , we have determined the 20-th percentile of standard errors in magnitude bins of @xmath53 mag straight line segments were then interpolated between these points to define a rejection criterion as a function of magnitude .
this has been done for each filter individually to account for the different characteristics of each bandpass and the different observing conditions for each dataset .
it is important to note that we do not automatically choose a lower boundary for brighter stars .
a less strict boundary for hb stars has been applied so as to obtain a larger , more complete sample in this evolutionary stage and thereby to better to define the observed zero - age horizontal branch ( zahb ) luminosity , which is an important distance and age indicator for the cluster .
furthermore , some pixels in the brightest stars occasionally exceed the saturation limit of the detector , which increases their observed photometric standard error . to produce a clear rgb sequence extending as far as possible towards brighter magnitudes
, we applied a less strict photometric rejection limit for these stars , as well . since the distribution of frames over the cluster field has an important impact on the final photometric standard error of a star
a large number of frames per observing field decreases the photometric error of stars therein the final hq - sample is not distributed homogeneously over the field , but favours areas that have been observed most frequently ( see figure [ hq_sample ] ) .
a significant fraction of field stars can be a problem in the analysis of gc cmds , since they are not located on the evolutionary sequence of the population .
therefore , they can smear the actual sequence and , in the worst case , lead to misinterpretations of observed features . by investigating our cmds of the whole cluster sample and comparing the total number of stars to those which
clearly do not follow the common cluster sequence , it is clear that the fraction of field stars lying in the line of sight to the object is very small for the field of view of our observations . from this ratio
, we infer the fraction of field stars to be @xmath54 .
therefore we assume that field stars do not affect our photometric analysis .
rr lyrae stars have been identified by their coordinates from the most recently updated database of c. clement ( @xcite ; 2011 edition ) . only cross - identifications with a coordinate deviation of
less than @xmath55 and only detections lying @xmath56 around the zahb @xmath57-band magnitude have been considered to be true rr lyrae identifications . by these criteria ,
35 rr lyrae stars were identified in our sample , which are plotted in figure [ hq cmds ] as black circles . except for a few outliers ,
these stars fall very well in the predicted instability strip . for all rr lyrae stars ,
we adopt the allframe magnitudes from our photometry and not the mean magnitudes . in our final hq - sample
, the photometry reaches to well below the main sequence turn - off ( msto ) to a magnitude of @xmath58 in @xmath57 and @xmath59 in j. notably , it reaches below the ms knee visible only in nir photometry which is an interesting feature for the determination of the distances , and hence ages , of globular clusters ( see @xcite ) . at the luminous end ,
the photometry is limited by the linearity limit of the nir data which lies at @xmath60 in @xmath57 and @xmath61 for the @xmath23 filter .
@xmath62 . the apparent larger spread in colour showed by the nir filter combination
is only caused by the smaller colour range covered by this cmd .
the black bars indicate the length and direction of the reddening vector for each filter combination assuming @xmath63 and a standard reddening law with @xmath64.,scaledwidth=50.0% ]
in our photometry of m4 , both the hb and the rgb sequences are poorly defined ( seen as a large scatter in magnitude on the hb and a large scatter in colour along the rgb ) compared to the well defined subgiant branch ( sgb ) and ms . additionally , in cmds with a combination of optical and nir filters , the colour scatter of the ms increases with luminosity towards the to . considering only the photometric uncertainty , one would expect the result to be the opposite .
furthermore , the scatter on the hb is much stronger in @xmath57 than it is in the @xmath23-band , suggesting that those effects are mainly caused by differential reddening instead of photometric errors .
generally , differential reddening induces the stars to scatter much more around the actual sequence than would be expected from their photometric uncertainty .
this additional scatter , however , is not uniformly evident throughout the gc fiducial sequence . in fact , it depends on the angle between the sequence and the reddening vector .
regions of the cmd where this angle is large show a larger scatter than regions where the sequence and the reddening vector are almost parallel .
the direction of the reddening vector depends upon the actual filters that are used and the extinction relation between these filters , defined by the reddening law . in figure [ hq cmds ]
we show the reddening vector for each individual filter combination using a standard reddening law with @xmath64 for purpose of illustration .
note , that for m4 we expect a slightly different orientation of the reddening vector .
the general idea for determining and correcting spatially differential reddening across the face of the cluster is to calculate for a sample of reference stars the distance from a fiducial sequence in a cmd as measured along the reddening vector . in theory
, this displacement depends only upon the individual reddening of the star and is therefore directly correlated with its spatial location in the cluster field .
now , the local reddening value defined by these reference stars can be provisionally applied to all program stars in their immediate neighbourhood ( see e.g. , @xcite , @xcite ) .
such a correction for spatially differential effects in the photometry potentially includes a first - order correction for poor flat - fielding or spatially varying psfs within different frames as well , in case they were not modelled perfectly during the data reduction process ( as is likely the case for our photometry ) .
our strategy for correcting differential reddening effects in m4 is a method of closest neighbours , " where the reddening displacement for each program star is determined as the median value among the closest 10 to 30 neighbours for which the residuals have been determined . with this approach ,
we are able to assign a reddening value on a star - by - star basis , allowing for any arbitrary dust geometry .
the resolution is automatically adjusted to the number density in a given area , determined by the neighbour with the largest distance which is used . obviously , this method smooths over , and is therefore insensitive to , reddening variations on any angular scale smaller than 36 times the typical angular separation between stars . in general , we adopt the method described by @xcite and implement only a few small changes . in order to obtain the maximal discrimination between the effects of differential reddening and those due to photometric error , we use a @xmath57 vs. @xmath65 cmd , which besides the large dynamic range of this colour compared to its uncertainty also provides another important advantage over other filter combinations . in general
, the direction of the reddening vector in the cmd depends upon the value of @xmath0 that describes the reddening law .
this filter combination , however , is highly insensitive to the reddening law due to the fact that the extinction in @xmath24 barely changes with the dust type .
therefore the effect on both axes is dominated by the change in @xmath57 and almost cancels out , with the result that we are able to perform the reddening correction without knowing @xmath0 precisely at this point . to define the angle of the reddening vector
, we use @xmath66 , assuming @xmath67 , but a standard assumption of 3.1 would only change the angle of the vector by less then 0.5 degree . in addition to a reference star sample on the main sequence , we select a second sample along the rgb which is smaller in number but shows stronger sensitivity to spatial differential reddening because it is more nearly orthogonal to the reddening vector .
since our data have low photometric uncertainty only for specific regions in the field of m4 , we exclude only stars with especially large uncertainties to be able to cover the majority of the surface area ( see table [ reference criteria ] ) . | we have used a combination of broad - band near - infrared and optical johnson - cousins photometry to study the dust properties in the line of sight to the galactic globular cluster m4 .
we have investigated the reddening effects in terms of absolute strength and variation across the cluster field , as well as the shape of the reddening law defined by the type of dust .
all three aspects had been poorly defined for this system and , consequently , there has been controversy about the absolute distance to this globular cluster , which is closest to the sun . here
, we determine the ratio of absolute to selective extinction ( @xmath0 ) in the line of sight towards m4 , which is known to be a useful indicator for the type of dust and therefore characterizes the applicable reddening law .
our method is independent of age assumptions and appears to be significantly more precise and accurate than previous approaches .
we obtain @xmath1 ( random error ) for the dust in the line of sight to m4 for our set of filters .
that corresponds to a dust - type parameter @xmath2 in the @xcite reddening law . with this value , the distance to m4
is found to be @xmath3 , corresponding to a true distance modulus of @xmath4 ( random error ) .
a reddening map for m4 has been created , which reveals a spatial differential reddening of @xmath5 across the field within @xmath6 around the cluster centre ; this is about 50% of the total mean reddening , which we have determined to be @xmath7 . in order to provide accurate zero points for the extinction coefficients of our photometric filters
, we investigated the impact of stellar parameters such as temperature , surface gravity and metallicity on the extinction properties and the necessary corrections in different bandpasses .
using both synthetic atlas9 spectra and observed spectral energy distributions , we found similar sized effects for the range of temperature and surface gravity typical of globular cluster stars : each causes a change of about @xmath8 in the necessary correction factor for each filter combination .
interestingly , variations in the metallicity cause effects of the same order when the assumed value is changed from the solar metallicity ( @xmath9}=0.0 $ ] ) to @xmath9=--2.5}$ ] .
the systematic differences between the reddening corrections for a typical main - sequence turnoff star in a metal poor globular cluster and a vega - like star are even stronger ( @xmath10 ) .
we compared the results from synthetic spectra to those obtained with observed spectral energy distributions and found significant differences for temperatures lower than @xmath11 .
we have attributed these discrepancies to the inadequate treatment of some molecular bands in the @xmath12 filter within the atlas9 models .
fortunately , these differences do not affect the principal astrophysical conclusions in this study , which are based on stars hotter than 5000k . from our calculations ,
we provide extinction zero points for johnson - cousins and 2mass filters , spanning a wide range of stellar parameters and dust types .
these extinction tables are suited for accurate , object - specific extinction corrections . |
we call a collection of congruent spheres in @xmath0-dimensional euclidean space @xmath1 a hard - sphere packing if no two spheres overlap . the study of the structure and macroscopic properties of hard spheres in physical dimensions ( @xmath14 or @xmath15 ) has a rich history , dating back to at least the work of boltzmann @xcite .
hard - sphere packings have been used to model a variety of systems , including liquids @xcite , amorphous and granular media @xcite , and crystals @xcite .
there has been resurgent interest in hard - sphere packings in dimensions greater than three in both the physical and mathematical sciences .
for example , it is known that the optimal way of sending digital signals over noisy channels corresponds to the densest sphere packing in a high dimensional space @xcite .
these error - correcting " codes underlie a variety of systems in digital communications and storage , including compact disks , cell phones and the internet .
physicists have studied hard - sphere packings in high dimensions to gain insight into ground and glassy states of matter as well as phase behavior in lower dimensions @xcite .
the determination of the densest packings in arbitrary dimension is a problem of long - standing interest in discrete geometry @xcite . the _ packing density _ or simply density @xmath16 of a sphere packing is the fraction of space @xmath1 covered by the spheres , i.e. , @xmath17 where @xmath18 is the number density , @xmath19 is the volume of a @xmath0-dimensional sphere of radius @xmath20 , and @xmath21 is the gamma function .
we call @xmath22 the _ maximal density _ , where the supremum is taken over all packings in @xmath1 .
the sphere packing problem seeks to answer the following question : among all packings of congruent spheres , what is the maximal packing density @xmath23 , i.e. , largest fraction of @xmath1 covered by the spheres , and what are the corresponding arrangements of the spheres @xcite ? for @xmath24 and @xmath15 , the optimal solutions are known .
for @xmath25 , the densest known packings of congruent spheres are bravais lattice packings @xcite , but in sufficiently large dimensions the optimal packings are likely to be non - bravais lattice packings .
upper and lower bounds on the maximal density @xmath23 exist in all dimensions @xcite .
for example , minkowski @xcite proved that the maximal density @xmath26 among all bravais lattice packings for @xmath27 satisfies the lower bound @xmath28 where @xmath29 is the riemann zeta function .
it is seen that for large values of @xmath0 , the asymptotic behavior of the _ nonconstructive _ minkowski lower bound is controlled by @xmath30 .
note that the density of a _ saturated _ packing of congruent spheres in @xmath1 for all @xmath0 satisfies @xmath31 which has the same dominant exponential term as ( [ mink ] ) .
a saturated packing of congruent spheres of unit diameter and density @xmath16 in @xmath32 has the property that each point in space lies within a unit distance from the center of some sphere . in the large - dimensional limit
, kabatiansky and levenshtein @xcite showed that the maximal density is bounded from above according to the asymptotic upper bound @xmath33 the present paper is motivated by some recent work on disordered sphere - packings in high dimensions @xcite . in ref .
@xcite , we introduced a generalization of the well - known random sequential addition ( rsa ) process for hard spheres in @xmath0-dimensional euclidean space @xmath1 .
this model can be viewed as a special thinning " of a poisson point process such that the subset of points at the end of the thinning process corresponds to a sphere packing .
one obvious rule is to retain a test sphere at time @xmath34 only if it does not overlap a sphere that was successfully added to the packing at an earlier time .
this criterion defines the standard rsa process in @xmath1 @xcite , which generates a homogeneous and isotropic sphere packing in @xmath1 with a time - dependent density @xmath35 . in the limit @xmath36 , the rsa process corresponds to a saturated packing with a maximal or _
density @xmath37 @xcite . in one dimension ,
the rsa process is commonly known as the car parking problem " , which re ' nyi showed has a saturation density @xmath38 @xcite . for @xmath39 , an exact determination of @xmath40
is not possible , but estimates for it have been obtained via computer experiments in two dimensions ( circular disks ) @xcite and three dimensions ( spheres ) @xcite .
however , estimates of the saturation density @xmath40 in higher dimensions have heretofore not been obtained .
another thinning criterion retains a test sphere centered at position @xmath41 at time @xmath34 if no other test sphere is within a unit radial distance from @xmath41 for the time interval @xmath42 prior to @xmath34 , where @xmath43 is a positive constant in the interval @xmath44 $ ] .
this packing is a subset of the rsa packing , which we call the generalized rsa process .
note that when @xmath45 , the standard rsa process is recovered , and when @xmath46 , we obtain the ghost " rsa process @xcite , which is amenable to exact analysis . in particular , we showed that the @xmath47-particle correlation function @xmath48 for the ghost rsa packing can be obtained analytically for any @xmath47 , all allowable densities and in any dimension .
this represents the first exactly solvable disordered sphere - packing model in arbitrary dimension . for statistically homogeneous packings in @xmath1 ,
these correlation functions are defined so that @xmath49 is proportional to the probability density for simultaneously finding @xmath47 particles at locations @xmath50 within the system , where @xmath18 is the number density .
thus , in a packing without long - range order , each @xmath51 approaches unity when all particle positions become widely separated within @xmath1 , indicating no spatial correlations . the fact that the maximal density @xmath52 of the ghost rsa packing implies that there may be disordered sphere packings in sufficiently high @xmath0 whose density exceeds minkowski s lower bound ( [ mink ] ) . indeed , in ref .
@xcite , a conjectural lower bound on the density of disordered sphere packings @xcite was employed to provide the putative exponential improvement on minkowski s 100-year - old bound .
the asymptotic behavior of the conjectural lower bound is controlled by @xmath53 .
these results suggest that the densest packings in sufficiently high dimensions may be disordered rather than periodic , implying the existence of disordered classical ground states for some continuous potentials .
in addition , a _
principle for disordered packings was identified @xcite , which states that _ unconstrained _ correlations in disordered sphere packings vanish asymptotically in high dimensions and that the @xmath51 for any @xmath54 can be inferred entirely from a knowledge of the number density @xmath18 and the pair correlation function @xmath55 . at first glace
, one might be tempted to conclude that the decorrelation principle is an expected mean - field " behavior , which is not the case .
for example , it is well known that in some spin systems correlations vanish in the limit @xmath56 and the system approaches the mean - field behavior . while this notion is meaningful for spin systems with attractive interactions , it is not for hard - core systems .
the latter is characterized by a total potential energy that is either zero or infinite , and thus can not be characterized by a mean field .
furthermore , mean - field theories are limited to equilibrium considerations , and thus do not distinguish between constrained " and unconstrained " correlations that arise in non - equilibrium packings of which there an infinite number of distinct ensembles .
the decorrelation principle is a statement about any disordered packing , equilibrium or not .
for example , contact delta functions ( constrained correlations ) are an important attribute of non - equilibrium jammed disordered packings and have no analog in equilibrium lattice models of any dimension .
finally , the decorrelation principle arises from the fact that the kabatiansky - levenshtein asymptotic upper bound on the maximal packing density ( [ kab ] ) implies that @xmath16 must go to zero at least as fast as @xmath57 for large @xmath0 and therefore , unconstrained spatial correlations between spheres are expected to vanish , i.e. , statistical independence is established @xcite .
there is no counterpart of the kabatiansky - levenshtein bound in mean - field theories . motivated by these recent results ,
we study the structural properties of standard rsa packings ( @xmath45 for the generalized rsa process ) in the infinite - time or saturation limit , generated via computer simulations , for the first six euclidean space dimensions ( @xmath58 ) . the algorithm is checked by reproducing some known results for @xmath24 and @xmath15 @xcite .
although we know that the saturation density @xmath40 is bounded from below by @xmath30 @xcite , the manner in which @xmath40 scales with density is not known for @xmath59 .
one objective of this paper is to answer this question .
another aim is determine the corresponding pair correlation functions and structure factors in order to ascertain whether decorrelations can be observed as the space dimension increases up to six .
a byproduct of our high - dimensional analysis is a relatively sharp lower bound on the saturation density of rsa packings for any @xmath0 .
although a palsti - like conjecture ( the saturation density in @xmath1 is equal to that of the one - dimensional problem raised to the @xmath0th power ) is exact for ghost rsa packings , we rigorously prove that this conjecture can not be true for standard rsa packings . in appendix
a , we obtain kissing ( contact ) number statistics for saturated rsa configurations on the surface of a @xmath0-dimensional sphere for dimensions @xmath12 and compare to the maximal kissing numbers in these dimensions . in appendix b , we determine the structure factor exactly for ghost rsa packings and show that its distance from hyperuniformity " @xcite increases as the space dimension increases , approaching a constant asymptotic value .
here we collect some known asymptotic results for the standard rsa process for @xmath0-dimensional hard spheres .
henceforth , we call @xmath60 the saturation ( infinite - time ) limit of the density . in his numerical study of rsa hard disks , feder @xcite postulated that the asymptotic coverage in the long - time limit for @xmath0-dimensional hard spheres follows the algebraic behavior @xmath61 where @xmath62 represents a dimensionless time .
theoretical arguments supporting feder s law ( [ feder ] ) have been put forth by pomeau @xcite and swendsen @xcite . not surprisingly , the saturation limit is approached more slowly as the space dimension increases . moreover , similar arguments lead to the conclusion that the pair correlation function @xmath63 at the saturation limit possesses a logarithmic singularity as the dimensionless radial distance @xmath64 for spheres of diameter @xmath65 approaches the contact value , independent of dimension @xcite , i.e. , @xmath66 boyer _ et al .
_ @xcite also showed that the pair correlation function for @xmath67 has _ super - exponential _ decay .
specifically , they found that at any finite time @xmath62 or density @xmath16 , @xmath68 thus , @xmath69 is a short - ranged function at any density .
this super - exponential decay of the pair correlation function persists in higher dimensions as well . as we will discuss in section iv
, this rapid decay of @xmath70 has implications for the analytic properties of the structure factor @xmath11 .
in what follows , we describe an efficient procedure to generate rsa packings in the saturation limit as well as the methods used to compute structural information , such as the density , pair correlation function , structure factor and cumulative coordination number .
we present a computationally fast method to generate rsa configurations of hard spheres in the saturation limit in the thermodynamic limit .
periodic boundary conditions are applied to a hypercubic fundamental cell of side length @xmath71 and volume @xmath72 .
spheres of diameter @xmath65 are placed randomly and sequentially inside the fundamental cell , which is periodically replicated to fill all of @xmath0-dimensional euclidean space @xmath1 , until the saturation limit is achieved . in order to speed up the computation , we attempt to add a particle only in the available space rather than wasting computational time in attempting to add particles anywhere in the fundamental cell @xcite .
this requires keeping track of the time - dependent available space , which is the space exterior to the union of the _ exclusion _ spheres of radius @xmath65 centered at each successfully added sphere at any particular time .
this is done by tessellating the hypercubic fundamental cell into smaller , disjoint hypercubic voxels , " which have side length between @xmath73 and @xmath74 , depending upon the dimension . at the start of the simulation ,
all voxels are declared accessible to particle placement .
there are two stages involved to determine the time - dependent available space .
a coarse estimation of the available space is used in the first stage , which is refined in the second stage . in the first stage
, a particle is successfully added to the simulation box , provided that it does not overlap any existing particle . to avoid checking for nonoverlaps with every successfully added particle to the simulation box
, we employ a neighbor list " @xcite , which amounts to checking within a local neighborhood of the attempted particle placement . for each successfully added particle ,
all voxels located within the largest possible inscribed hypercube centered at the exclusion sphere of radius @xmath65 that are fully occupied by the particle are declared to be part of the unavailable space .
the voxels outside this hypercube but within the exclusion sphere may be partially filled . in the initial stages ,
such partially filled voxels are declared to be part of the available space .
we call these accessible voxels .
if there have been at least one million unsuccessful placement attempts since the last accepted particle placement , we move to the second stage to refine our determination of the available space . in particular
, we determine whether each remaining accessible voxel from the first stage can accommodate a particle center by a random search of each accessible voxel .
after about 1000 random placement attempts , a particle is either added to a particular voxel or this voxel is declared to be part of the unavailable space .
this search is carried out for all other accessible voxels .
the simulation terminates when all voxels are unavailable for particle addition in the second stage .
this two - stage procedure enables us to generate rsa packings that are saturated or nearly saturated .
generating truly saturated rsa packings becomes increasingly difficult as the space dimension increases , as the asymptotic relation ( [ feder ] ) indicates . at any instant of time
@xmath62 , the number @xmath75 of added particles for a particular configuration is known and the density @xmath76 is computed from relation ( [ phi ] ) with @xmath77 .
we call @xmath78 the stopping " density , i.e. , the density at the time @xmath79 when the simulation is terminated
. the system size @xmath80 is sufficiently large so as produce a histogram for @xmath81 or @xmath4 that is gaussian distributed .
although we do not present the full distribution of densities here , we do report the associated standard errors . in order to estimate the true saturation density @xmath4 , the volume of the available space @xmath82 as a function of dimensionless time @xmath62 is recorded in the very late stages , namely , for the last ten particles added . the saturation density @xmath4
is estimated from this late - stage data by plotting @xmath76 versus @xmath83 [ cf .
( [ feder ] ) ] and extrapolating to the infinite - time limit .
however , to perform the extrapolation properly the time increment between each particle addition can not be taken to be uniform but instead must increase with increasing time in order to account for the fact that we only attempt to add particles in the available space . in the very late stages , this time increment @xmath84 is given by @xmath85 the stopping density @xmath86 always bounds the saturation density @xmath87 from below , but as we will soon see , @xmath86 is very nearly equal to the saturation density @xmath4 .
we obtain the pair correlation function @xmath63 at the nearly - saturated stopping density @xmath86 for a specific configuration by generating a histogram of the average number of particle centers @xmath88 contained in a concentric shell of finite thickness @xmath89 at radial distance @xmath64 from an arbitrary reference particle center @xcite .
the radial distance @xmath64 is defined as halfway between the inner radius ( @xmath90 ) and the outer radius ( @xmath91 ) of each shell .
the shell thickness is termed the bin width .
let @xmath92 represent the accumulated pairs of particles for the entire system placed in bin @xmath93 associated with a radial distance @xmath64 . by definition
, @xmath92 must be an even integer .
then @xmath94 where @xmath95 is the number of particles in the fundamental cell . in general , the pair correlation ( or radial distribution ) function is defined as @xmath96 where @xmath97 is the volume of the @xmath0-dimensional shell , given by @xmath98 \hspace{5pt } , \label{equation3}\ ] ] @xmath18 is the number density @xmath99 , and @xmath100 is the volume of a @xmath0-dimensional sphere of radius @xmath64 as shown earlier .
we compute ensemble - averaged pair correlation functions by binning up to a maximum distance of @xmath101 for each realization of the ensemble and then averaging over all ensemble members . away from contact
, we employ a bin width of @xmath102 .
near contact , we use a finer bin width of @xmath103 in order to accurately capture the logarithmic divergence of @xmath70 as the contact value is approached .
another quantity of interest is the cumulative coordination number @xmath104 , which gives the average number of sphere centers within a distance @xmath64 from a given sphere center .
it is related to the pair correlation function @xmath105 as follows : @xmath106 where @xmath107 is the @xmath0-dimensional surface area of a sphere of radius @xmath64 @xcite .
finally , we also compute the structure factor @xmath108 , which provides a measure of the density fluctuations at a particular wave vector @xmath109 and is defined by the relation @xmath110 where @xmath111 is the fourier transform of the total correlation function @xmath112 .
when the total correlation in @xmath1 depends on the wavenumber @xmath113 , then the structure factor @xmath11 in @xmath1 for any space dimension @xmath0 is given by @xcite @xmath114 where @xmath115 is the bessel function of order @xmath116 .
the expression ( [ factor2 ] ) provides a means for computing the structure factor by fourier transforming the real - space total correlation function in @xmath1 .
if one is interested in the large - wavelength ( small @xmath93 ) behavior , however , the large @xmath64 behavior of @xmath117 must be known with high precision .
even for relatively large simulation cells , it is difficult to access this large-@xmath64 asymptotic behavior . in such instances , it is better to compute the structure factor directly from the collective density variables , i.e. , @xmath118 where @xmath119 are the collective density variables , angular brackets denote an ensemble average , and @xmath109 are the wave vectors appropriate for the periodic cell of volume @xmath120 . for the hypercubic fundamental cell of side length @xmath71 considered here ,
the @xmath0-dimensional wave vectors are given by @xmath121 where @xmath122 ( @xmath123 ) are the integers .
thus , the smallest positive wave vector that one can measure has magnitude @xmath124 . for small to intermediate values of @xmath93
, we will employ the direct method , while for intermediate to large values of @xmath93 , we will use both the direct and indirect method [ i.e. , we calculate @xmath11 using ( [ factor2 ] ) ] .
the saturation density @xmath4 for each of the first six space dimensions was determined by considering 100 - 1,000 realizations and several different system sizes , as discussed in section iii.b .
table [ tablei ] summarizes our results for the saturation density for the largest systems and the associated standard error .
included in the table is the stopping density @xmath86 , relative system volume @xmath125 for the largest system , where @xmath126 is the volume of a hypersphere , and the total number of configurations @xmath127 .
the results for @xmath24 and @xmath15 agree well with known results for these dimensions @xcite .
we see that the stopping density @xmath86 is very nearly equal to the saturation density @xmath4 for all dimensions , except for @xmath67 where these two quantities are identical . for @xmath67 ,
no extrapolation was required since we can ensure that the packings were truly saturated in this instance . .
[ tablei ] the computed saturation density @xmath87 and associated standard error for the first six space dimensions . included in the table is the stopping density @xmath86 , relative system volume @xmath125 , and the total number of configurations @xmath127 . [
cols="^,^,^,^,^",options="header " , ] table [ table4 ] provides kissing number statistics for saturated rsa configurations for dimensions @xmath12 . for @xmath14 ,
only two values of @xmath128 are allowed : 4 and 5 . any kissing number equal to 3 or less
is prohibited because such configurations are not saturated . on the other hand ,
a kissing number of 6 has a probability of zero of occurring by random sequential addition because the configuration corresponding to this optimal kissing number is unique .
we find that the average kissing number is approximately equal to 4.5 .
for @xmath129 , the average kissing number @xmath130 . the fact that the maximal kissing number configurations ( @xmath131 ) in @xmath132 are nonunique implies that a configuration of 12 spheres has a positive ( albeit small ) probability of occurring via an rsa process . nonetheless , we were not able to observe such a configuration in a total of @xmath133 configurations .
the smallest observed kissing number for @xmath129 was six , which presumably is the smallest number required for a saturated packing .
the minimal kissing number configurations for saturation are related to the following problem : how can @xmath47 points be distributed on a unit sphere such that they maximize the minimum distance between any pair of points ? for six points , the solution to his problem is well known : they should be placed at the vertices of an inscribed regular octahedron . since the minimum angular separation between any pair of points in this highly symmetric case is 90 degrees , the associated kissing number configuration is saturated .
note that highest kissing number of 18 reported for @xmath134 is substantially smaller than the maximal kissing number @xmath135 .
apparently , achieving kissing numbers that approach those of the optimal highly symmetric , rigid 24-cell configuration by a random sequential addition is effectively impossible .
therefore , it is plausible that the possible configurations corresponding to kissing numbers of 19 - 23 are also characterized by high degree of symmetry ( and possibly rigidity ) based upon the absence of such kissing numbers .
the fact that the average kissing number for @xmath136 is substantially lower than the highest known kissing number of 40 is presumably related to the high symmetries required to achieve high @xmath137 values in this dimension .
our data for the average rsa kissing number over the range of considered dimensions is fit very well by the following quadratic expression in @xmath0 : @xmath138 where @xmath139 , @xmath140 and @xmath141 , and the correlation coefficient is 0.9999 .
the data and this quadratic fit function are depicted in figure [ kiss ] .
if this expression persisted for large @xmath0 , it would predict that the average rsa kissing number asymptotically grows as @xmath142 .
for ghost rsa packings of spheres of diameter @xmath65 in @xmath1 , the pair correlation function in the infinite - time limit is given exactly for any space dimension @xmath0 by the following expression @xcite : @xmath143 where @xmath144 is the unit step function , equal to zero for @xmath145 and unity for @xmath146 , and @xmath147 is the intersection volume of two spheres of radius @xmath65 whose centers are separated by the distance @xmath64 divided by the volume of a sphere of radius @xmath65 .
expressions for the _ scaled _ intersection volume @xmath147 for any @xmath0 are known exactly ; see refs . @xcite and @xcite for two different representations .
the scaled intersection volume @xmath147 takes its maximum value of unity at @xmath148 and monotonically decreases with increasing @xmath64 such that it is nonzero for @xmath149 , i.e. , it has compact support . the corresponding total correlation @xmath150 is given by @xmath151 we see that that @xmath117 can be written as a sum of two contributions : the pure step function contribution @xmath152 , which has support for @xmath153 , and a contribution involving @xmath147 , which has support @xmath154 .
substitution of ( [ h2 ] ) into ( [ factor2 ] ) yields the structure factor to be @xmath155 where @xmath156 is the structure factor for the step function contribution @xmath152 , and @xmath157 is the contribution to @xmath11 in excess to the structure factor for the step function .
here we have used the fact that the infinite - time density is @xmath158 . for odd dimensions
, @xmath159 can be obtained explicitly in terms of sine , cosine , sine integral and cosine integral functions .
we do not explicitly present these expressions here but instead plot @xmath11 , defined by ( [ combined ] ) , for various dimensions in fig .
[ ghosts ] . for @xmath160 and @xmath161 , @xmath162
is given by 0.150728 , 0.290134 and 0.452217 , respectively , and therefore not only is the ghost rsa packing not hyperuniform , as expected , but its distance from hyperuniformity increases as the space dimension @xmath0 increases , asymptotically approaching the value of 1/2 .
this should be contrasted with the standard rsa packing in @xmath1 , which we have shown is nearly hyperuniform . in the limit @xmath56
, the excess contribution to the structure factor has the limiting form @xmath163 ^ 2= \frac{1}{2}[1-s_{sf}(k)]^2 . \label{asympt}\ ] ] the resulting structure factor in this asymptotic limit is given by @xmath164 , \qquad d \rightarrow \infty.\end{aligned}\ ] ] it has the following small-@xmath93 expansion : @xmath165 the asymptotic result ( [ asympt ] ) is easily obtained by utilizing the fact that in the limit @xmath56 , @xmath166 \rightarrow \alpha_2(r;d)/2 $ ] @xcite .
substitution of this result into the general relation ( [ excess ] ) and recognizing that the lower limit @xmath65 of this integral can be replaced by 0 in the limit @xmath167 yields the asymptotic form @xmath168 where @xmath169 denotes the fourier transform of @xmath147 and @xmath170 is the volume of a sphere of radius @xmath65 [ cf .
( [ v(r ) ] ) ] .
the quantity @xmath169 is known explicitly in any dimension @xcite and substitution of this result into ( [ int ] ) immediately yields ( [ asympt ] ) .
note that the infinite - time rsa limit has also been called the jammed " limit , but , for general packing problems , the term _ jammed _ has more recently come to signify a certain degree of mechanical rigidity due to the existence of an interparticle contact network .
the reader interested in the latter sense of jamming is referred to s. torquato and f. h. stillinger , j. phys .
b * 105 * , 11849 ( 2001 ) ; s. torquato , a. donev , and f. h. stillinger , int .
j. solids structures * 40 * , 7143 ( 2003 ) ; and a. donev , s. torquato , f. h. stillinger and r. connelly , j. comp .
* 197 * , 139 ( 2004 ) . in ref .
@xcite , a _
disordered packing _ in @xmath32 is defined to be a packing in which the pair correlation function @xmath55 decays to its long - range value of unity faster than @xmath171 for some @xmath172 . | sphere packings in high dimensions have been the subject of recent theoretical interest .
employing numerical and theoretical methods , we investigate the structural characteristics of random sequential addition ( rsa ) of congruent spheres in @xmath0-dimensional euclidean space @xmath1 in the infinite - time or saturation limit for the first six space dimensions ( @xmath2 ) .
specifically , we determine the saturation density , pair correlation function , cumulative coordination number and the structure factor in each of these dimensions . we find that for @xmath3 , the saturation density @xmath4 scales with dimension as @xmath5 , where @xmath6 and @xmath7 .
we also show analytically that the same density scaling persists in the high - dimensional limit , albeit with different coefficients .
a byproduct of this high - dimensional analysis is a relatively sharp lower bound on the saturation density for any @xmath0 given by @xmath8 , where @xmath9 $ ] is the structure factor at @xmath10 ( i.e. , infinite - wavelength number variance ) in the high - dimensional limit . we prove rigorously that a palsti - like conjecture ( the saturation density in @xmath1 is equal to that of the one - dimensional problem raised to the @xmath0th power ) can not be true for rsa hyperspheres .
we demonstrate that the structure factor @xmath11 must be analytic at @xmath10 and that rsa packings for @xmath2 are nearly hyperuniform . "
consistent with the recent decorrelation principle , " we find that pair correlations markedly diminish as the space dimension increases up to six .
we also obtain kissing ( contact ) number statistics for saturated rsa configurations on the surface of a @xmath0-dimensional sphere for dimensions @xmath12 and compare to the maximal kissing numbers in these dimensions .
we determine the structure factor exactly for the related ghost " rsa packing in @xmath1 and show that its distance from hyperuniformity " increases as the space dimension increases , approaching a constant asymptotic value of @xmath13 .
our work has implications for the possible existence of disordered classical ground states for some continuous potentials in sufficiently high dimensions . |
modulational instabilities ( mis ) have a time - honored history in nonlinear wave equations . their occurrences span areas ranging from fluid dynamics @xcite ( where they are usually referred to as the benjamin - feir instability ) and nonlinear optics @xcite to plasma physics @xcite .
while earlier manifestations of such instabilities were studied in continuum systems @xcite , in the last decade the role of the mi in the dynamics of discrete systems has emerged . in particular , the mi was analyzed in the context of the discrete nonlinear schrdinger equation @xcite , a ubiquitous nonlinear - lattice dynamical model @xcite .
more recently , it was studied in the context of weakly interacting trapped bose - einstein condensates ( becs ) , where the analytical predictions based on the discrete model @xcite were found to be in agreement with the experiment @xcite ( see also the recent works @xcite , and recent reviews in @xcite ) .
additionally , the development of discrete nonlinear optics ( based on nonlinear waveguide arrays ) has recently provided the first experimental observation of the mi in the latter class of systems @xcite .
an extension of these recent works , which is relevant to both becs in the presence of an optical - lattice potential @xcite and nonlinear optics in photorefractive crystals @xcite , is the case of multi - component discrete fields .
these can correspond to a mixture of two different atomic species , or different spin states of the same atom , in becs @xcite , or to light waves carried by different polarizations or different wavelengths in optical systems @xcite .
the corresponding two - component model is based on two coupled discrete nonlinear schrdinger ( dnls ) equations , @xmath0this is a discrete analog of the well - known model describing nonlinear interactions of the above - mentioned light waves through self - phase - modulation and cross - phase - modulation ( xpm ) @xcite . in optics ,
( [ ceq1 ] ) describe an array of optical waveguides , the evolution variable being the propagation distance @xmath1 ( rather than time @xmath2 ) .
the choice of the nonlinear coefficients in the optical models is limited to the combinations @xmath3 for orthogonal linear polarizations , and @xmath4 for circular polarizations or different carrier wavelengths .
in becs , the coefficients @xmath5 in eqs .
( [ ceq1 ] ) , are related to the three scattering lengths @xmath6 which account for collisions between atoms belonging to the same ( @xmath7 ) or different ( @xmath6 , @xmath8 ) species ; in that case , @xmath9 ( @xmath10 ) corresponds to the repulsive ( attractive ) interaction between the atoms .
the linear coupling between the components , which is accounted for by the coefficient @xmath11 in eqs .
( [ ceq1 ] ) , is relevant in optics for a case of circular polarizations in an array of optical fibers with deformed ( non - circular ) cores , or for linear polarizations in an array of twisted fibers @xcite . on the other hand , in the becs context
, @xmath11 represents the rabi frequency of transitions between two different spin states in a resonant microwave field @xcite . stimulated by the experimental relevance of the mi in discrete coupled systems , the aim of the present work is to develop a systematic study of the instability in the two - component dynamical lattices .
we give an analytical derivation of the mi criteria for the case of both the nonlinear and linear coupling between the components . as a result of the analysis
, we also find a novel domain - wall ( dw ) stationary pattern in the case of the linear coupling . the presentation is structured as follows : in the following sections we derive plane - wave solutions and analyze their stability , corroborating it with a numerical analysis of the stability intervals .
we do this for the model with the linear coupling in section ii and for the one with the purely nonlinear coupling in section iii . in section iv
, we examine dw states in the linearly coupled lattices .
finally , the results and findings are summarized in section v.
we look for plane - wave solutions in the form @xmath12 , \quad j=1,2 .
\label{plane}\ ] ] the linear coupling imposes the restrictions @xmath13 and @xmath14 . inserting eq .
( [ plane ] ) into eqs .
( [ ceq1 ] ) yields @xmath15 in the particular , but physically relevant , symmetric case , with @xmath16 and @xmath17 , it follows from here that the amplitudes @xmath18 obey an equation @xmath19\left(a_{1}^{2}-a_{2}^{2}\right)=0 $ ] , hence either of the following two relations must then be satisfied : @xmath20@xmath21 in the case of eq .
( [ plane1 ] ) , a nonzero solution has @xmath22 .
it exists with @xmath23 , provided @xmath24 , and with @xmath25 , if @xmath26 .
when @xmath27 , one obtains solutions of the form @xmath28 with arbitrary @xmath29 and @xmath30 . on the other hand , from eq .
( [ plane2 ] ) , one finds that @xmath31under the restriction that this expression must be positive .
when @xmath32 , solutions of this type exist as long as @xmath33 s_{11 } > 0 $ ] ( as the term under the square root is smaller in magnitude than the one outside ) and the argument of the square root in ( [ tip2 ] ) is non - negative .
the first condition implies @xmath34 , for @xmath35 , and @xmath36 , for @xmath37 . in the case
@xmath38 , eq . ( [ tip2 ] ) takes the simpler form @xmath39 .
% \label{int}\ ] ] for both @xmath38 and @xmath40 , it is necessary to impose the condition @xmath41 for the solutions to be real .
we note that these relations are similar to those derived in ref .
@xcite , where linearly and nonlinearly coupled systems of continuum nls equation were considered .
it is also worth noting that the existence of two _ distinct _ uniform states with a fixed product from eq .
( [ plane2 ] ) suggests a possibility of a domain - wall ( dw ) solution in the model with the linear coupling .
dw solutions in nonlinearly coupled discrete nonlinear schrdinger equations were examined in ref .
@xcite ( following an analogy with the continuum ones of ref .
@xcite ) . however , the present case is different , as both uniform states have non - vanishing amplitudes in both components , and , as seen from eq .
( [ plane2 ] ) , the dws may exist only if @xmath32 .
this possibility is examined in more detail in section iv . to examine the stability of the plane waves , we substitute @xmath42\exp [ i(qn-\omega t ) ] \label{ceq1a}\ ] ] into eqs .
( [ ceq1 ] ) , to obtain a system of two coupled linearized equations for the perturbations @xmath43 .
furthermore , assuming a general solution of the above - mentioned system of the form @xmath44 where @xmath45 and @xmath46 are the wavenumber and frequency of perturbation , we arrive at a set of four homogeneous equations for @xmath47 , @xmath48 , @xmath49 and @xmath50 .
the latter have a nontrivial solution if @xmath45 and @xmath46 satisfy the dispersion relation @xmath51 \\ & & \times \left [ ( \omega -2d_{2}\sin { q}\sin { q})^{2}-\left ( 2d_{2}\cos { q}(\cos { q}-1)+c\frac{a_{1}}{a_{2}}\right ) \left ( 2d_{2}\cos { q}(\cos { q}-1)+c\frac{a_{1}}{a_{2}}-2s_{22}a_{2}^{2}\right ) \right ] \\ & & -2c(2s_{12}a_{1}a_{2}+c)(\omega -2d_{1}\sin { q}\sin { q})(\omega -2d_{2}\sin { q}\sin { q})-c^{2}\left ( 2d_{1}\cos { q}(\cos { q}-1)+c\frac{a_{2}}{a_{1}}-2s_{11}a_{1}^{2}\right ) \\ & & \times \left ( 2d_{2}\cos { q}(\cos { q}-1)+c\frac{a_{1}}{a_{2}}-2s_{22}a_{2}^{2}\right ) -(2s_{12}a_{1}a_{2}+c)^{2}\end{aligned}\]]@xmath52 = 0 .
\label{ceq4}\ ] ] note that in the absence of coupling , i.e. , @xmath53 , we obtain a known relation @xcite @xmath54 which gives the mi condition if the right - hand side becomes negative , i.e. , @xmath55 we will now consider the particular case @xmath56 , the general case being technically tractable but too involved .
then , the dispersion relation ( [ ceq4 ] ) takes the from @xmath57where @xmath58@xmath59 .\end{aligned}\ ] ] it immediately follows from eq .
( [ ceq5 ] ) that , to avoid the mi , both solutions for @xmath60 should be positive .
taking into account the binomial nature of the equation , it is concluded that the spatially homogeneous solution is unstable if either the sum @xmath61 or the product @xmath62 of the solutions is negative : @xmath63 to proceed further , we may fix the value of the perturbation wavenumber , to investigate in what parameter region it would give rise to the mi .
we illustrate this approach in figs .
[ rfig1 ] and [ rfig2 ] , in which we fix @xmath64 and @xmath65 and vary @xmath11 and @xmath66 ( the coefficients of the linear and xpm coupling ) , to examine their effect on the stability interval . from eq .
( [ k_r ] ) we see that for these values of the parameters the modulational unstable region is @xmath67 .
it can be inferred from the figures that @xmath11 may widen the mi interval by decreasing its lower edge . on the other hand , @xmath66 has a more complex effect : while making the instability interval larger by increasing its upper edge ( until it reaches @xmath68 ) , it may also open mi bands within the initially modulationally stable region ( see , e.g. , the lower panel of fig . [ rfig2 ] ) . , the cases of @xmath69 ( top panel ; unstable for @xmath70 ) , @xmath71 ( middle panel ; unstable for @xmath72 ) and @xmath73 ( bottom panel ; unstable for @xmath74 ) .
the solid line shows the sum @xmath75 and the dashed line the product @xmath76 of the solutions of eq .
( [ ceq5 ] ) .
the instability takes place in intervals of the wavenumber @xmath77 of the unperturbed plane - wave solution where either @xmath75 or @xmath76 ( or both ) are negative.,width=384,height=288 ] but fixing @xmath71 and varying @xmath66 .
the figure shows the cases of @xmath78 ( top panel ; unstable for @xmath79 ) , @xmath80 ( middle panel ; unstable for @xmath81 ) , and @xmath82 ( bottom panel ; unstable for @xmath83 and @xmath84 ) .
, width=384,height=288 ] for reasons of completeness , we also illustrate nonlinear development of the mi in the coupled nonlinear lattices in some typical examples . in particular ,
the role of the mi in generating large - amplitude excitations in the presence of the linear coupling only ( @xmath38 ) is illustrated in the left panel of fig .
[ rfig2a ] , for @xmath73 . in the right panel
of the figure , the mi in the presence of both linear and nonlinear coupling ( @xmath73 , @xmath78 ) is shown .
it is readily observed that the nonlinear coupling enhances the instability growth rate and , hence , the mi sets in earlier . )
contour plots of the squared absolute value of one of the two components of the solution ( top ) and temporal evolution of the maximum of the solution s amplitude ( bottom ) , @xmath73 , @xmath38 and @xmath85 .
right panels : same as the left ones but for @xmath73 , @xmath78 and @xmath85.,title="fig:",width=288,height=288 ] ) contour plots of the squared absolute value of one of the two components of the solution ( top ) and temporal evolution of the maximum of the solution s amplitude ( bottom ) , @xmath73 , @xmath38 and @xmath85 .
right panels : same as the left ones but for @xmath73 , @xmath78 and @xmath85.,title="fig:",width=288,height=288 ]
a physically relevant case that we also wish to consider here is the one with only the nonlinear coupling present , i.e. , @xmath69 .
the corresponding dispersion relations read @xmath86 to study the stability of the plane waves in this case , we use eq .
( [ ceq1a ] ) as before , and obtain the following dispersion relation for the perturbation wavenumber and frequency @xmath87 \nonumber \\ \times & & \left [ ( \omega -2d_{2}\sin { q}\sin { q_{2}})^{2}-2d_{2}\cos { q_{2}}(\cos { q}-1)\left ( 2d_{2}\cos { q_{2}}(\cos { q}-1)-2s_{22}a_{2}^{2}\right ) \right ] \nonumber \\ - & & 4(2s_{12}a_{1}a_{2})^{2}d_{1}d_{2}\cos { q_{1}}\cos { q_{2}}(\cos { q}-1)^{2}=0 .
\label{purenonlin}\end{aligned}\ ] ] when @xmath38 , the known result ( [ k_r ] ) for the one - component case is easily retrieved .
if we let @xmath56 and @xmath13 , then eq .
( [ purenonlin ] ) is simplified as follows : @xmath88 where @xmath89 we will follow the lines of the analysis outlined above for the case of the linear coupling . examining the product and the sum of the roots of eq .
( [ ceq9 ] ) for @xmath90 , which now read @xmath91 and @xmath92 , we arrive at the following mi conditions : @xmath93 the latter can be rewritten , respectively , as : @xmath94with @xmath95 . it is readily observed , in this case as well , that the coupling between the two components tends to expand the band of the mi wavenumbers with respect to the single - component case .
the effect of the variation of @xmath66 in this case , for fixed @xmath65 and @xmath64 , is shown in fig .
[ rfig3 ] .
notice also that eqs .
( [ ceqn5 ] ) and ( [ ceqn6 ] ) suggest multi - component generalizations of the mi criteria . but setting @xmath69 and varying @xmath66 .
the figure shows the cases of @xmath78 ( top panel ; unstable for @xmath96 ) , @xmath80 ( middle panel ; unstable for @xmath81 ) , and @xmath82 ( bottom panel ; unstable for @xmath97).,width=384,height=288 ] when @xmath98 , @xmath99 and @xmath100 , the mi conditions given by eqs .
( [ ceqn5])-([ceqn6 ] ) can be written in a compact form , @xmath101 from eq .
( [ ceqn5a ] ) , we obtain that ( for @xmath102 ) @xmath103 while similarly from eq .
( [ ceqn6a ] ) , it follows that @xmath104 examining the latter relations in more detail , we conclude that , for @xmath105 , the region of unstable wavenumbers is @xmath106 , while , for @xmath107 , eq . ( [ t2 ] ) contains the interval of unstable @xmath77 .
this is illustrated for the case with @xmath108 as a function of @xmath66 in fig .
[ fig_add ] . for the modulational instability vs. @xmath66 , as found from eqs .
( [ t1 ] ) and ( [ t2 ] ) ( for the case of @xmath108).,width=384,height=288 ] a typical example of the simulated development of the mi in the case of purely nonlinear coupling is shown in fig .
[ rfig3a ] for @xmath78 and @xmath109 . , but for @xmath78 and @xmath69 ( purely nonlinear coupling ) ; @xmath85.,width=384,height=288 ]
the expressions ( [ tip2 ] ) for the amplitudes of the plane - wave solutions suggest a novel possibility : focusing more specifically on the so - called anti - continuum limit of @xmath110 ( we use @xmath111 ) and following the path of @xcite , we can construct _ domain - wall _ ( dw ) solutions that connect the homogeneous state given by eq .
( [ tip2 ] ) with its `` conjugate '' state of @xmath112 .
such a solution can then be continued for finite coupling , to examine its spatial profile and dynamical stability .
an example of such a solution , for the case where the nonlinear coupling is absent , is given in fig .
[ rfig4 ] .
it is observed that the solution is stable for @xmath113 , and it becomes unstable due to a cascade of oscillatory instabilities ( through the corresponding eigenvalue quartets ) for larger values of @xmath114 .
, as a function of the inter - site coupling constant @xmath114 using continuation from the _ anti - continuum limit _ , @xmath110 .
the bottom subplot shows the most unstable eigenvalue of small perturbations around the solution as a function of @xmath114 .
the dw becomes unstable for @xmath115 .
the profiles of the solution ( top subplots ) and the respective linear - stability eigenvalues ( bottom subplots ) are shown in the right panel for @xmath116 ( stable ; left subplots ) and @xmath117 ( unstable ; right subplots ) .
the parameters are @xmath38 , @xmath71 , @xmath118.,title="fig:",width=288,height=288 ] , as a function of the inter - site coupling constant @xmath114 using continuation from the _ anti - continuum limit _ , @xmath110 .
the bottom subplot shows the most unstable eigenvalue of small perturbations around the solution as a function of @xmath114 .
the dw becomes unstable for @xmath115 .
the profiles of the solution ( top subplots ) and the respective linear - stability eigenvalues ( bottom subplots ) are shown in the right panel for @xmath116 ( stable ; left subplots ) and @xmath117 ( unstable ; right subplots ) .
the parameters are @xmath38 , @xmath71 , @xmath118.,title="fig:",width=288,height=288 ] we have also examined the evolution of the dws when they are unstable .
a typical result is displayed in fig .
[ rfig4a ] .
it is seen from the bottom panel , which shows the time evolution of the solution s maximum amplitude , that growth of the oscillatory instability eventually destroys the configuration , through lattice turbulence .
this apparently chaotic evolution can be attributed to the mixing of a large number of unstable eigenmodes .
the dynamics remain extremely complex despite the eventual saturation of the instability . .
the middle panel shows the same for the second field .
the bottom panel displays the time evolution of the amplitude ( the spatial maximum ) of the first field.,width=384,height=288 ] the increase of the nonlinear coupling constant @xmath66 reduces the stability window of these solutions . in particular , the case of @xmath119 is shown in fig .
[ rfig5 ] in which , the stability window has shrunk to @xmath120 . , but for @xmath119 .
the instability sets in for @xmath121.,title="fig:",width=288,height=288 ] , but for @xmath119 .
the instability sets in for @xmath121.,title="fig:",width=288,height=288 ] one can also consider a modified dw , where an extra site between the two domains has equal or opposite amplitudes of the two fields .
we have checked that such solutions are always unstable ( due to the presence of real eigenvalue pairs ) , for all values of @xmath114 . still more unstable ( with a larger number of unstable eigenvalues ) are more sophisticated dw patterns , with additional intermediate sites inserted between the two domains .
in this work , we have examined the extension of the modulational instability ( mi ) concept to the case of multiple - component discrete fields .
we have shown , in a systematic way , how to formulate the linear mi equations and how to extract the mi criteria .
we have also followed the dynamical evolution of the instability by means of direct simulations , and have identified the effects of the linear and nonlinear couplings on the range of modulationally unstable wavenumbers . in particular , we have demonstrated that the joint action of the two couplings may give rise to noteworthy features , such as opening of new mi bands on the wavenumber scale .
additionally , the identification of a pair of conjugate uniform solutions in the two - component model has prompted us to examine domain - wall ( dw)solutions between such states .
we were able to demonstrate that the dws can be linearly stable , provided that the inter - site coupling in the lattice is sufficiently weak . from our results
, it is clear that multi - component lattice models have a rich phenomenology , which is a natural addition to that of single - component ones
. it would be interesting to observe the predicted features in experimental settings , including weakly coupled becs and photonic - crystal nonlinear media .
_ acknowledgements .
_ zr and pgk gratefully acknowledge the hospitality of the center of nonlinear studies of the los alamos national laboratory where part of this work was performed .
pgk also acknowledges the support of nsf - dms-0204585 , nsf - career and the eppley foundation for research .
djf acknowledges support of the special research account of university of athens .
the work of bam was supported , in a part , by the israel science foundation through the grant no .
research at los alamos is performed under the auspices of the us - doe . for recent reviews
see , e.g. , p. g. kevrekidis , k. .
rasmussen , and a. r. bishop , int .
b , * 15 * , 2833 ( 2001 ) ; j. c. eilbeck and m. johansson, _ proc . of the 3rd conf
localization & energy transfer in nonlinear systems ( june 17 - 21 2002 , san lorenzo de el escorial madrid ) _ , ed . l. vzquez _
( world scientific , new jersey , 2003 ) , p. 44
( arxiv : nlin.ps/0211049 ) .
r. j. ballagh , k. burnett , and t. f. scott , phys .
* 78 * , 1607 ( 1997 ) ; j. williams , r. walser , j. cooper , e. cornell , and m. holland , phys .
a * 59 * , r31 ( 1999 ) ; p. hberg and s. stenholm , phys .
a * 59 * , 3890 ( 1999 ) .
see , e.g. , c. j. myatt _ et al .
_ , phys.rev .
* 78 * , 586 ( 1997 ) ; d. s. hall _ et al .
_ , phys . rev
. lett . * 81 * , 1539 ( 1998 ) ; d. m.stamper-kurn _ et al .
_ , phys .
lett.*80 * , 2027 ( 1998 ) ; g. modugno _ et al .
_ , science * 294 * , 1320 ( 2001 ) ; m. mudrich _ et al .
_ , phys .
lett . * 88 * , 253001 ( 2002 ) ; m. trippenbach _ et al_. , j. phys .
b * 33 * , 4017 ( 2000 ) ; s. coen and m. haelterman , phys .
* 87 * , 140401 ( 2001 ) ; p. hberg and l. santos , phys .
* 86 * , 2918 ( 2001 ) ; th .
busch and j. r. anglin , phys .
* 87 * , 010401 ( 2001 ) . | we consider a system of two discrete nonlinear schrdinger equations , coupled by nonlinear and linear terms .
for various physically relevant cases , we derive a modulational instability criterion for plane - wave solutions . we also find and examine domain - wall solutions in the model with the linear coupling . |
resonant atomic nonlinearities involving higher excited states enable strong multiphoton interactions , facilitating efficient quantum frequency conversion @xcite , photon pair generation @xcite and single - photon level nonlinearities @xcite .
multiphoton processes also enable selective and precise control of atomic states @xcite .
in particular , non - degenerate four - wave mixing ( 4wm ) in a four - level configuration offers extensive possibilities , such as frequency conversion @xcite or coherent blue light ( cbl ) generation @xcite . in volume atomic vapor , as in bulk nonlinear crystals , phase matching is required to obtain efficient four - wave mixing
. in particular , spatial arrangement of laser beams must enable the phase - matching condition @xcite .
since atomic vapors are dilute in comparison to solid matter , a unity refractive index for all contributing beams is assumed when considering the phase - matching condition , thus making light propagation effects trivial .
this approach corresponds to a simple single - atom perspective and all phase relations throughout the ensemble are neglected .
these assumptions however , are not correct in the regime where four - wave mixing is most efficient - close to atomic resonances and in a dense atomic vapor .
multiple experiments encounter effects that arise due to phase matching : these effects manifest themselves in quite non - obvious ways , such as shifting of wave - mixing resonances @xcite or strong modifications of measured spectra @xcite
. proper identification of important contributions is arduous and although simulations provide reasonable agreement @xcite , little insight is gained .
only a handful of works consider spatial @xcite , dispersive or wavevector mismatching effects in atomic processes such as electromagnetically induced transparency @xcite or four - wave mixing in a double-@xmath1 configuration used to generate twin - beams @xcite . here
we show that a vast majority of observed effects may be described by a simple model of spatial propagation and phase matching .
we use our results to engineer the spatial configuration of the beams to find that four - wave mixing may be enhanced if phase matching is taken into account .
we also demonstrate direct effects of rapidly - varying refractive index on wave - mixing resonance position , shape and light emission direction .
we take one of its hyperfine components with @xmath2 and for the highest excited state @xmath3 we take its hyperfine component with @xmath4 . for the intermediate state @xmath5
we consider all its hyperfine components , that lead to multiple wave - mixing paths .
( b ) beam directions : three - dimensional sketch of the laser beams intersecting inside a cell filled with rubidium vapor .
laser beams marked with green and blue colors have wavelength of 780 nm .
laser beam marked with the red arrow ( 776 nm ) propagates in the opposite direction .
the four - wave mixing signal is marked with the orange dotted line .
the signal is emitted roughly in direction determined by transverse phase matching , where the sum of transverse components of wavevectors @xmath6 is equal to the respective @xmath7 sum . due to this condition the four points in the transverse @xmath8-plane ,
corresponding to the four beams , form a parallelogram .
( c ) central part of the experimental setup . emitted four - wave mixing signal is filtered using a half - wave plate ( @xmath9 ) , a polarizer and an interference filter and then registered by with an avalanche photodiode or a ccd camera situated in the far field with respect to the rubidium vapor cell .
[ fig : schemat],scaledwidth=100.0% ] the four - wave mixing signal is generated by the nonlinear polarization created in the atomic medium by three crossing beams ( see fig .
[ fig : schemat](a ) ) .
we consider a general situation where three incident , non - degenerate collimated beams ( no . 1 , 2 and 4 ) generate a new signal field ( beam 3 ) . as presented in fig .
[ fig : schemat](c ) the fields are arranged in the double - ladder configuration being nearly - resonant to atomic transitions .
beams 1 and 2 take part in two - photon absorption .
beams 3 and 4 take part in two - photon emission , where beam 4 stimulates the emission to facilitate theoretical description of the process , let us first consider propagation of a monochromatic signal in a nonlinear atomic medium with polarization @xmath10 .
approximate equation for the electric field envelope @xmath11 of this signal has the following well - known form ( see ref .
@xcite ) : @xmath12 where @xmath13 , @xmath14 is the linear polarizability , @xmath15 is emitted light frequency and @xmath16 . in the transverse fourier domain ( @xmath17 ) the equation takes on the following form : @xmath18 where we have used the exponential ansatz for rapidly - varying longitudinal phase dependance @xmath19 and tilde indicates the fourier transform of a respective field . in the experiment
we start with no four - wave mixing signal at the input @xmath20 , so the above equation may be integrated to yield the signal field amplitude at the exit of the atomic medium : @xmath21 high macroscopic amplitude at the output is generated only when the spatial oscillations in @xmath22-direction inside the integral are slow in comparison with the medium length @xmath23 , and thus the correct phase matching is required .
diameters of incoming beams equal to 400 @xmath24 m .
panel ( a ) portrays the gaussian intensity distribution without taking into account the longitudinal phase - matching condition .
panel ( b ) shows the value of the phase - matching factor @xmath25 .
the black circle corresponds to a region where the longitudinal phase matching is perfect .
variations of the refractive index induced by changing laser detuning cause the radius of this circle to change as well , which in turn impacts the direction of emission.[fig : symul],scaledwidth=100.0% ] the incoming laser beams indexed by @xmath26 are assumed to be gaussian with equal diameters of @xmath27 .
their frequencies @xmath28 determine their wavevector lengths @xmath29 , with @xmath30 being the refractive index for respective beam .
they propagate at small angles @xmath31 with respect to the @xmath22-axis , so it is convenient to replace wavevectors with angles @xmath32 , @xmath33 .
the desired final result is the angular distribution of emitted four - wave mixing signal @xmath34 and its total intensity @xmath35 let us assume that the amplitude of incident beam @xmath36 is maximal along the @xmath22-axis , neglecting shifts of beams centers : @xmath37 corresponding nonlinear atomic polarization generated in such a system is : @xmath38 where @xmath39 and the resulting polarization wavevector @xmath40 this wavevector is oriented at an angle @xmath41 to the @xmath22-axis , so transverse coordinates are given by @xmath42 inserting the expression for nonlinear polarization [ eq .
( [ eq : p4wm ] ) ] with the wavevector given by eq .
( [ eq : duzefi ] ) to the integral [ eq . ( [ eq : obw ] ) ] , we obtain the final distribution of the four - wave mixing signal : @xmath43 @xmath44 the above formula for the distribution consists of two factors : ( a ) the gaussian factor that depends only on transverse variables and is the result which is obtained if we neglect the longitudinal phase matching and ( b ) the @xmath45 factor that represents the longitudinal phase - mismatch .
figure [ fig : symul ] shows exemplary theoretical intensity distribution of emitted beam without taking into account the longitudinal phase - matching condition ( i.e. the gaussian factor , at the left ) and the ring - shaped phase - matching factor ( at the right ) .
shape of the four - wave mixing intensity distribution depends on angles and wavevector lengths of incoming beams .
the gaussian factor is virtually independent of frequencies .
indeed , small changes of wavevector lengths shift the gaussian factor by much less than its width @xmath46 , so they can be neglected .
however , since @xmath47 , we need very small @xmath48 [ see eq .
( [ eq : delta ] ) ] for efficient phase matching .
equation ( [ eq : delta ] ) can be rewritten as @xmath49 which for @xmath50 describes the cone of perfect phase matching . in the transverse plane , the best phase - matching
is achieved on the ring with radius @xmath51 with angular width of approximately @xmath52 .
we now consider changes of the radius @xmath51 as a function of incident and emitted beams wavevectors .
the derivative over @xmath53 is @xmath54 a typical value we find for @xmath51 in the experiment is of the order of @xmath55 rad , so derivative is of the order of @xmath56 m@xmath57 .
therefore , change of refractive index of the order of @xmath58 causes change of @xmath51 of the order of @xmath59 rad , much more than phase - matching ring width . by analogy
we can calculate derivative over incident beam angles @xmath60 : @xmath61 consequently , unless the spatial configuration is properly designed , the overlap between phase - matching ring and gaussian factor from eq .
( [ eq : amplituda ] ) could potentially decrease with any change of @xmath62 and resulting change of the phase - matching ring radius @xmath51 .
we now consider linear and nonlinear polarizabilities essential to description of the four - wave mixing process .
the third order polarizability can be obtained through perturbation chain for a specific case we consider in the experiment @xcite : @xmath63 where @xmath64 , @xmath65 is the decay rate , @xmath66 are dipole moments of transitions between respective states and @xmath67 is the atomic concentration .
additionally , since phase matching plays a crucial role , we take into account dispersion of refractive index .
the linear polarizability for beam 1 is : @xmath68 where @xmath69 is the rabi frequency for beam 2 .
the linear polarizability for beam 4 takes on analogous form . for beam 2
we have the following linear polarizability : @xmath70 and the polarizability for beam 3 will take on an analogous form as well . to complete our considerations we take the doppler broadening into account .
all , linear and nonlinear , polarizabilities should be averaged over the maxwell velocity distribution @xcite : @xmath71 @xmath72 where @xmath73 is the atomic mass , @xmath74 is boltzmann s constant and @xmath75 is temperature .
dominant contribution of the doppler broadening is present in single - photon terms , which we take into account as full voigt profiles in the above calculation .
however , there is also residual doppler broadening in the two - photon terms with @xmath76 . since calculation of this term is time - consuming while the broadening is small , we decided to replace it by additional natural broadening , which is of quite different shape , but gives similar and consistent results
essential components and ideas of the experiment are presented in fig . [
fig : schemat ] .
we use a double - ladder level configuration in @xmath77rb ( fig . [
fig : schemat ] ) .
the four - wave mixing signal is generated on the transition between @xmath3 and @xmath5 manifolds . to drive the process we use three lasers : one external - cavity diode laser ( ecdl , number 2 ) with wavelength of 776 nm and linewidth approx .
100 khz and two distributed feedback ( dfb ) laser diodes with wavelengths of 780 nm ( number 1 and 4 ) and linewidth of about 1 mhz . to minimize the two - photon doppler broadening laser beams 1 and 2
are arranged in the counter - propagating configuration .
the 776 nm laser is stabilized using a commercial wavemeter ( highfinesse ws7 ) .
one of the 780 nm lasers is locked at the vicinity of two - photon absorption peak using an auxiliary rubidium vapor cell placed in a tunable magnetic field where we crossed strong circularly polarized 776 nm laser beam and weak linearly polarized laser 1 beam at 780 nm . by measuring polarization rotation in circular basis
we could generate tunable locking signal , that controls laser 1 and lock the two lasers at @xmath78 .
more details of this method are described in ref .
using the fact that difference between frequencies of lasers 1 and 4 is of the order of several ghz , laser 4 is stabilized by beat - note measurement .
all laser beams intersect in the cell with warm rubidium vapors at natural abundance and no buffer gas .
figure [ fig : schemat](b ) depicts the central part of the setup and a telescope used to obtain beams intersecting at approx .
@xmath79 mrad angle and @xmath80 diameter of 400 @xmath24 m to eliminate influence of external magnetic fields the cell is placed inside a double @xmath24-metal shielding . additionally , the cell is heated using a bifilarly - wound coil to avoid stray magnetic field from the heater .
the generated four - wave mixing signal is separated from stray driving light using a half - wave plate with a polarizing beamsplitter and then using a band - pass interference filter tilted to transmit light at 776 nm .
the signal is registered by an avalanche photodiode ( apd ) or a ccd camera situated in the far field of the rubidium cell , allowing us to measure angular distribution of the emission .
a flip - mirror is used to select either the apd or the ccd camera .
for @xmath81 mhz and @xmath82 at @xmath83 .
dots correspond to the experimental result , while solid lines correspond to theoretical prediction .
[ fig : przesuwanie],scaledwidth=75.0% ] in the vicinity of the two - photon resonance .
dots correspond to experimental result for @xmath84 mhz and @xmath85 mhz , while solid lines correspond to the theoretical prediction .
vertical dotted lines mark maxima of the four - wave mixing signal intensity and two - photon absorption .
the frequency shift between the two maxima is marked as @xmath86 , while the maximum intensity of the four - wave mixing signal attained at the resonance is marked as @xmath87 .[fig : przes],scaledwidth=75.0% ] ( blue curves , left axis ) and corresponding frequency shifts @xmath86 ( red curves , right axis ) for a set of @xmath88 and @xmath89 detuning . panels ( a ) and ( b ) correspond to experimental and theoretical results , respectively .
subsequent curves were measured / calculated for different @xmath88 detunings in 500 @xmath90mhz steps starting from @xmath91 mhz on the left.[fig : przes_cale],scaledwidth=75.0% ] -axis . dots and crosses inside circles correspond to direction to the positive and negative values on the @xmath22-axis .
( b ) dependence of the four - wave mixing frequency shift @xmath86 in relation to the two - photon absorption resonance and ( c ) maximum intensity @xmath92 on @xmath89 detuning .
subsequent experimental points correspond to measurements at various detuning @xmath89 and angles as shown in panel ( a ) with constant @xmath93 dots correspond to experimental result , while solid lines correspond to theoretical prediction .
[ fig : kat],scaledwidth=75.0% ] in the following section we compare our experimental and theoretical results to demonstrate the importance of the phase - matching condition and finally engineer the optimal geometry for the process .
first , we observe the dependence of the average emission angle @xmath94 of the four - wave mixing signal on detuning @xmath89 , while remaining detunings are kept constant . in particular ,
driving beam 1 is far detuned from the resonance ( @xmath81 mhz ) , while being simultaneously kept in the two - photon resonance , i.e. @xmath82 . from the theoretical point of view , the phase - matching ring , as described in sec .
[ sec : teoryja ] , has finite width and overlaps with the gaussian factor from eq .
( [ eq : amplituda ] ) at either side .
for that reason , the emission angle @xmath94 changes as a function of detuning ( in this case @xmath89 ) . in the experiment ,
the full signal beam profile at the far field is measured with the ccd camera and the average emission angle is inferred from the image after electronic background subtraction . in the theoretical framework
this corresponds to the angle calculated as : @xmath95 measurement results and theoretical predictions results are presented in fig .
[ fig : przesuwanie ] .
we observe a strong correspondence between the two , which confirms that the reason of dependence between emission direction and lasers detunings is indeed dispersion and resulting change in phase matching .
as far as the change of emission angle is perhaps the most direct consequence of dispersion and phase matching , another significant effect is the shift in optimum two - photon detuning @xmath76 for the four - wave mixing signal generation in relation to two - photon absorption maximum .
this effect , sometimes also visible as a strong modification of spectrum , has been observed in several previous works @xcite , but its origin remained unexplained .
simple theory , which neglects dispersion of refractive indices , predicts that maximum intensity of measured signal is exactly at the two - photon absorption maximum , as the third order polarizability @xmath96 and the two - photon absorption coefficient proportional to @xmath97 depends on the two - photon detuning @xmath76 in the very same way [ see eqs .
( [ eq : chi3 ] ) and ( [ eq : chi2 ] ) ] . in our model not only large polarizability @xmath96 is required , but also overlap between the phase - matching ring and gaussian factor from eq .
( [ eq : amplituda ] ) .
the latter condition is not necessarily met at @xmath82 . from this reasoning ,
the frequency shift @xmath86 of maximum follows .
an exemplary plot of the two - photon absorption and the four - wave mixing signals as a function of the two - photon detuning are presented in fig .
[ fig : przes ] .
note a significant shift @xmath86 between the extrema of the two signals , marked with dashed lines .
imperfect fit to theory is due to residual two - photon doppler broadening , accounted for in numerical calculations as a broadened lorentzian instead of a full voigt profile .
the shift was measured for multiple values of detunings @xmath88 and @xmath89 .
experimental results , together with theoretical prediction , are presented in fig .
[ fig : przes_cale ] . for different detunings
the system exhibits different dispersion profiles , and thus we observe different frequency shifts @xmath98 .
they confirm that dispersion and phase matching accurately explain dependence between the four - wave mixing signal intensity and laser detunings .
note that the dispersion in the vicinity of the two - photon resonance changes with the optical depth of the medium ( or atom number density ) and consequently quite strongly depends on temperature of the atomic vapor cell @xcite .
finally , we confirm that as originally intended the geometry of the experiment may be changed to tailor the four - wave mixing signal properties . from theory
it follows that the overlap between the phase - matching ring and the gaussian factor decreases with any change of incident beams tilts @xmath99 , @xmath100 and @xmath101 [ see eq .
( [ eq : dthdfi ] ) ] .
however , this overlap can fixed by adjusting detunings . as an example
, we study how the maximum of the four - wave mixing signal changes as a function of @xmath89 detuning if we tilt the incident beam 3 .
we performed the measurement for three different tilts presented in fig .
[ fig : kat](a ) . figures [ fig : kat](b ) and [ fig : kat](c ) portray obtained experimental and theoretical results for the frequency shift @xmath86 and maximum intensity @xmath92 , respectively .
we observe that the maximum of the four - wave mixing signal changes quite significantly .
note , that without taking into account the phase - matching condition we only obtain @xmath102 dependance for the field amplitude [ as in eq .
[ eq : chi3 ] ] , which would clearly give a completely incorrect theoretical prediction in this case . instead , the proper choice of geometry is critical when laser 4 is tuned closer or further from its respective single - photon resonance .
these results demonstrate that our theoretical model may be used to precisely predict the behavior of the four - wave mixing signal , explaining a variety of intricacies .
in particular , we may predict the optimum geometry for a desired set of detunings , or vice - versa .
we have shown that a simple model of four - wave mixing in an atomic medium that neglects propagation effects could be easily extended by considering propagation equation with nonlinear polarization to account for phase matching . in the spatial fourier domain
, the phase - matching condition implies that only some component wavevectors of the drive beams lead to effective wave - mixing .
moreover , due to dispersion , the phase - matching conditions strongly depend on laser detunings .
this approach allowed us to explain some phenomena which were unexplainable earlier , like the frequency shift of the four - wave mixing signal maximum in relation to two - photon absorption maximum @xcite .
the most direct consequence of dispersion seems to be the change of the four - wave mixing signal emission angle as a function of laser detunings .
these results show that our theoretical description is much more robust than an atomic model neglecting the influence of dispersion on phase matching in atomic medium and allows precise predictions of the four - wave mixing signal behavior .
moreover , we have shown that our model facilitates proper choice of geometry and consequently stronger signal may be obtained . apart from providing a recipe for engineering of effective interaction , we believe that our results may also help study other subtle effects in four - wave mixing @xcite , where multiple shift effects might contribute to the total frequency shift . finally , we note that our approach is generally valid for a variety of systems - one needs to simply supply appropriate expressions for linear and nonlinear susceptibilities .
in particular , effects treated in this work become even more critical in high - density ensembles , that facilitate strong light - matter interactions due to collective effects @xcite , in waveguides that support modes with various longitudinal wavevectors @xcite , in cavity - enhanced processes @xcite or finally in schemes that rely on phase - matching control , such as multimode quantum memories @xcite .
if the four - wave mixing signal is weak as in quantum light - atom interfaces @xcite or processes involving rydberg states @xcite proper choice of geometry may be even more critical and yield a difference between having a strong signal or not obtaining the four - wave mixing signal at all .
the project was financed by the polish ministry of science and higher education `` diamentowy grant '' project no . di2013 011943 and by the national science centre ( poland ) grants no .
2015/16/s / st2/00424 and 2015/17/d / st2/03471 .
we acknowledge generous support of t. stacewicz , k. banaszek and r. apkiewicz as well as careful proofreading of the manuscript by m. dbrowski . | multiphoton processes in dense atomic vapors such as four - wave mixing or coherent blue light generation are typically viewed from single - atom perspective . here
we study the surprisingly important effect of phase matching near two - photon resonances that arises due to spatial extent of the atomic medium within which the multiphoton process occurs .
the non - unit refractive index of the atomic vapor may inhibit generation of light in nonlinear processes , significantly shift the efficiency maxima in frequencies and redirect emitted beam .
we present these effects on an example of four - wave mixing in dense rubidium vapors in a double - ladder configuration . by deriving a simple theory that takes into account essential spatial properties of the process
, we give precise predictions and confirm their validity in the experiment .
the model allows us to improve on the geometry of the experiment and engineer more efficient four - wave mixing .
10 a. g. radnaev , y. o. dudin , r. zhao , h. h. jen , s. d. jenkins , a. kuzmich , and t. a. b. kennedy , `` a quantum memory with telecom - wavelength conversion , '' nature phys .
* 6 * , 894899 ( 2010 ) .
t. chanelire , d. matsukevich , s. jenkins , t. kennedy , m. chapman , and a. kuzmich , `` quantum telecommunication based on atomic cascade transitions , '' phys .
rev . lett . * 96 * , 093604 ( 2006 ) .
b. srivathsan , g. k. gulati , b. chng , g. maslennikov , d. matsukevich , and c. kurtsiefer , `` narrow band source of transform - limited photon pairs via four - wave mixing in a cold atomic ensemble , '' phys .
rev .
lett . *
111 * , 123602 ( 2013 ) .
r. t. willis , f. e. becerra , l. a. orozco , and s. l. rolston , `` photon statistics and polarization correlations at telecommunications wavelengths from a warm atomic ensemble . ''
opt .
express * 19 * , 1463214641 ( 2011 ) .
t. peyronel , o. firstenberg , q .- y .
liang , s. hofferberth , a. v. gorshkov , t. pohl , m. d. lukin , and v. vuleti , `` quantum nonlinear optics with single photons enabled by strongly interacting atoms , '' nature * 488 * , 5760 ( 2012 ) .
m. parniak , a. leszczyski , and w. wasilewski , `` coupling of four - wave mixing and raman scattering by ground - state atomic coherence , '' phys .
rev .
a * 93 * , 053821 ( 2016 ) .
b. huber , a. klle , and t. pfau , `` motion - induced signal revival in pulsed rydberg four - wave mixing beyond the frozen gas limit , '' phys .
rev . a * 90 * , 053806 ( 2014 ) .
m. parniak and w. wasilewski , `` interference and nonlinear properties of four - wave - mixing resonances in thermal vapor : analytical results and experimental verification , '' phys .
rev . a * 91 * , 023418 ( 2015 ) .
p.
s. donvalkar , v. venkataraman , s. clemmen , k. saha , and a. l. gaeta , `` frequency translation via four - wave mixing bragg scattering in rb filled photonic bandgap fibers . ''
opt .
lett . * 39 * , 15571560 ( 2014 ) .
r. t. willis , f. e. becerra , l. a. orozco , and s. l. rolston , `` four - wave mixing in the diamond configuration in an atomic vapor , '' phys . rev . a * 79
* , 033814 ( 2009 ) .
f. e. becerra , r. t. willis , s. l. rolston , and l. a. orozco , `` nondegenerate four - wave mixing in rubidium vapor : the diamond configuration , '' phys .
rev .
a * 78 * , 013834 ( 2008 ) .
e. brekke , j. o. day , and t. g. walker , `` four - wave mixing in ultracold atoms using intermediate rydberg states , '' phys .
rev .
a * 78 * , 063830 ( 2008 ) .
t. meijer , j. d. white , b. smeets , m. jeppesen , and r. e. scholten , `` blue five - level frequency - upconversion system in rubidium , '' opt
. lett . *
31 * , 10021004 ( 2006 ) .
j. f. sell , m. a. gearba , b. d. depaola , and r. j. knize , `` collimated blue and infrared beams generated by two - photon excitation in rb vapor , '' opt .
lett . * 39 * , 528531 ( 2014 ) .
a. vernier , s. franke - arnold , e. riis , and a. s. arnold , `` enhanced frequency up - conversion in rb vapor . ''
opt .
express * 18 * , 1702017026 ( 2010 ) .
a. s. zibrov , m. d. lukin , l. hollberg , and m. o. scully , `` efficient frequency up - conversion in resonant coherent media , '' phys .
rev .
a * 65 * , 051801 ( 2002 ) .
a. a. m. akulshin , r. j. r. mclean , a. i. sidorov , and p. hannaford , `` coherent and collimated blue light generated by four - wave mixing in rb vapour . ''
opt .
express * 17 * , 2286122870 ( 2009 ) .
r. w. boyd , _ nonlinear optics ( third edition ) _
( academic press , 2008 ) .
e. brekke and e. herman , `` frequency characteristics of far - detuned parametric four - wave mixing in rb . ''
. lett .
* 40 * , 56745677 ( 2015 ) .
z. wang , y. zhang , p. li , s. sang , c. yuan , h. zheng , c. li , and m. xiao , `` observation of polarization - controlled spatial splitting of four - wave mixing in a three - level atomic system , '' appl .
phys .
b * 104 * , 633638 ( 2011 ) .
j. han , t. vogt , m. manjappa , r. guo , m. kiffner , and w. li , `` lensing effect of electromagnetically induced transparency involving a rydberg state , '' phys . rev .
a * 92 * , 063824 ( 2015 ) .
v. bharti and a. wasan , `` complete wavelength mismatching effect in a doppler broadened @xmath0-type six - level eit atomic medium , '' opt .
commun . * 324 * , 238244 ( 2014 ) .
a. b. mirza and s. singh , `` wave - vector mismatch effects in electromagnetically induced transparency in @xmath0-type systems , '' phys .
rev .
a * 85 * , 053837 ( 2012 ) .
m. t. turnbull , p. g. petrov , c. s. embrey , a. m. marino , and v. boyer , `` role of the phase - matching condition in nondegenerate four - wave mixing in hot vapors for the generation of squeezed states of light , '' phys .
rev .
a * 88 * , 033845 ( 2013 ) .
a. b. mirza and s. singh , `` electromagnetically induced transparency and steady - state propagation characteristics in doppler broadened diamond systems , '' j. mod . opt .
* 62 * , 1626 ( 2015 ) .
m. parniak , a. leszczyski , and w. wasilewski , `` magneto - optical polarization rotation in a ladder - type atomic system for tunable offset locking , '' appl .
phys .
lett . * 108 * , 161103 ( 2016 ) .
n. r. de melo and s. s. vianna , `` frequency shift in three - photon resonant four - wave mixing by internal atom - field interaction , '' phys .
rev .
a * 92 * , 053830 ( 2015 ) . b. m. sparkes , j. bernu , m. hosseini , j. geng , q. glorieux , p. a. altin , p. k. lam , n. p. robins , and b. c. buchler , `` gradient echo memory in an ultra - high optical depth cold atomic ensemble , '' new j. phys .
* 15 * , 085027 ( 2013 ) .
r. f. offer , j. w. c. conway , e. riis , s. franke - arnold , and a. s. arnold , `` cavity - enhanced frequency up - conversion in rubidium vapor , '' opt .
lett . * 41 * , 21772180 ( 2016 ) .
m. mazelanik , m. dabrowski , and w. wasilewski , `` correlation steering in the angularly multimode raman atomic memory , '' opt .
express * 24 * , 2199522003 ( 2016 ) .
n. r. de melo and s. s. vianna , `` two - photon resonant forward four - wave mixing in rubidium vapor involving rydberg states , '' j. opt .
soc . am .
b * 31 * , 1735 - 1740 ( 2014 )
. w. c. magno , r. b. prandini , p. nussenzveig , and s. s. vianna , `` four - wave mixing with rydberg levels in rubidium vapor : observation of interference fringes , '' phys .
rev .
a * 63 * , 063406 ( 2001 ) . |
spin systems with cyclic exchange interactions have been receiving an increasing amount of attention in the past few years .
these interactions are known to be responsible for the properties of the triangular magnetic system composed of @xmath1he atoms absorbed on graphite surfaces.@xcite recently , it was demonstrated that a certain amount of four - spin exchange could explain the neutron - scattering experiments concerning high-@xmath2 compounds such as la@xmath3cuo@xmath4,@xcite , la@xmath5ca@xmath6cu@xmath7o@xmath8,@xcite , and la@xmath4sr@xmath9cu@xmath10o@xmath8.@xcite the latter two are spin - ladder compounds where charge fluctuations such as cyclic hopping processes modify the heisenberg spin hamiltonian by contributing a four - spin interaction term . in particular , it was found that these interactions substantially modify the spin triplon gaps and frustrate the formation of bound triplon states.@xcite on the theoretical side , in spite of the numerous numerical studies predicting rich phase diagrams,@xcite a number of important questions , concerning the type of spin orderings and quantum criticalities realized by the four - spin exchange , remain unsettled.@xcite in this paper we analyze the role of the cyclic four - spin exchange interactions in the symmetric diamond chain ( sdc ) which is one of the simplest spin models admitting multispin cyclic exchange interactions .
the model is defined by the hamiltonian ( see fig . [ dchain ] ) [ h ] h=_n=1^l h_n , n+1 , where h_n , n+1&=&j_1*s*_n ( _ n+_n+1 ) + j_n_n+1 + & + & j_*s_1*_n_n + k . here
@xmath11 , @xmath12 , and @xmath13 are spin-@xmath0 operators defined on the sites of the @xmath14-th elementary cell . in our model ,
the standard cyclic four - spin exchange interactions@xcite are slightly generalized by including its bilinear terms in the exchange parameters @xmath15 , @xmath16 , and @xmath17 . in
what follows the action and the energy are measured in the units of @xmath18 and @xmath16 , respectively .
the frustrated diamond chain@xcite as well as its various modifications , such as the distorted diamond chain@xcite and the so - called @xmath19 ferrimagnetic chain with ising and heisenberg spins@xcite , have already been discussed in the literature in relation to some quasi-1d magnetic materials.@xcite for the following analysis it is important to notice that the cyclic exchange interaction does not violate the local symmetry of the hamiltonian under the exchange of the pair of spins ( @xmath20 ) for each diamond in the sdc .
thus , in the important case of spin-@xmath0 operators @xmath11 and @xmath12 , the system is characterized by @xmath21 local good quantum numbers @xmath22 ( @xmath23 ) related to the composite spins @xmath24 : @xmath25 . using this local symmetry and the standard commutation relations for spin operators ,
the local hamiltonian @xmath26 can be represented in the compact form [ h2 ] h_n , n+1&=&_n + * s*_n ( _ n+_n+1 ) + j_n_n_n+1 + & + & \ { * s*_n_n , * s*_n_n+1 } .
here @xmath27 are fixed numbers ( @xmath28 or @xmath29 ) for every sector defined as a sequence of the local quantum numbers @xmath30 $ ] , @xmath31 , and @xmath32 is the anticommutator of two operators ( @xmath33 and @xmath34 ) . the symmetric diamond chain considered in the paper .
the arrows denote the cyclic four - spin exchange interaction controlled by the parameter @xmath35 . ,
width=264 ] in the parameter space where the ground state is characterized by @xmath36 ( @xmath37 ) , the second and fourth terms in eq .
( [ h2 ] ) vanish and the model is equivalent to the spin-@xmath0 heisenberg chain with an exchange parameter @xmath38 . on the other hand ,
if the ground state belongs to the sector @xmath39 ( @xmath37 ) , eq .
( [ h2 ] ) defines the generic model of an alternating mixed - spin chain with the simplest three - spin exchange interaction .
finally , if the ground state is in the sector [ @xmath40 , the system is reduced to some modification of the well - known orthogonal dimer chain.@xcite
it is instructive to begin with the phase diagram of a single diamond composed of the spins @xmath41 , @xmath42 , @xmath43 , and @xmath44 .
the diamond eigenstates ( see table 1 ) consist of two singlets ( @xmath45 and @xmath46 ) , three triplets ( @xmath47 , @xmath48 , and @xmath49 ; @xmath50 ) , and the quintet @xmath51 ( @xmath52 ) . in figure
[ ph_diagram ] , the phase boundaries between the single - diamond ground states ( @xmath45 , @xmath46 , @xmath53 , and @xmath54 ) are depicted by dashed lines . as discussed below ,
some of these lines coincide with the exact phase boundaries of the diamond chain , eq .
( [ h ] ) .
.[table]eigenvalues and eigenstates of a single diamond composed of the spins @xmath55 , @xmath56 , @xmath43 , and @xmath44 .
the eigenstates are classified according to the following good quantum numbers of the single - diamond cluster : the total diamond spin @xmath57 , its @xmath58 component @xmath59 , and the diagonal spins @xmath60 and @xmath61 [ @xmath62 and @xmath63 : @xmath64 , @xmath65 .
the symbols @xmath66 ( @xmath67 ) stand for the canonical basic states of the spin-1 operators , whereas @xmath68 denotes the singlet state of two spin-@xmath0 operators . for brevity , only the @xmath69 components of the triplet ( @xmath47 ) and quintet ( @xmath51 ) states are presented . [ cols="<,<,<,<,<,<",options="header " , ] for large enough values of the parameter @xmath17 , the off - chain spins form local dimers , @xmath71 so that the ground state of the model belongs to the sector @xmath70 $ ] .
thus , eq . ( [ h2 ] ) reduces to the form h_n , n+1=-(3/4)j_+(j+k/4)_n_n+1 .
this is the hamiltonian of a spin-1/2 heisenberg chain with the exchange constant @xmath38 .
the single - diamond line @xmath72 ( defined by @xmath73 ) coincides with the exact boundary between the fully polarized ferromagnetic phase fm1 ( @xmath74 ) and the critical spin - fluid phase sf ( @xmath75 ) .
phase diagram of the model , eq .
( [ h ] ) , in the parameter space ( @xmath76 ) for @xmath77 as obtained from the numerical diagonalization of periodic clusters containing up to @xmath78 unit cells .
the dashed lines denote the boundaries of the single - diamond phase diagram containing the states @xmath45 , @xmath46 , @xmath53 , and @xmath54 ( see table 1 ) .
the following abbreviations for the phases of eq .
( [ h ] ) are used : critical spin - fluid state ( sf ) , fully - polarized ferromagnetic state ( fm1 ) , ferrimagnetic state ( fm2 ) , tetramer - dimer states ( td1,td2 ) , dimerized singlet state ( dm ) , and another ferrimagnetic state ( fm3 ) .
, width=283 ] besides the well - documented collective modes , these phases exhibit additional single - particle modes ( related to the off - diagonal spins ) describing excited ( @xmath79 dimers , @xmath39 .
being eigenstates of the hamiltonian , eq . ( [ h ] ) , these excitations are completely localized . according to eq .
( [ h2 ] ) , an excited dimer ( say at @xmath80 ) is described by the hamiltonian [ hi ] h_i= c+(j+ ) _
n=1^l-1_n_n+1+h_i , where @xmath81 and @xmath82 reads [ hii ] h_i & = & + ( j - ) _ l_1+j_i * s*_l ( _ l + _ 1 ) + & + & \ { * s*_l_l , * s*_l_1 } , j_i=1 .
similar models , describing spin-@xmath83 impurities in spin-1/2 antiferromagnetic heisenberg chains , have been discussed in the literature.@xcite in particular , the above model may be considered as a special class of open chains with symmetric couplings of the end spins ( @xmath84 and @xmath85 ) to the external spin @xmath86 ( @xmath87 ) . turning to the region occupied by the sf phase
, renormalization - group arguments predict that symmetric perturbations to the open chain are , at most , marginal .
thus , the marginally relevant perturbation @xmath88 ( @xmath89 ) is expected to renormalize to @xmath90 in the sf phase .
this corresponds to a fixed point where the end spins @xmath84 and @xmath85 are effectively removed from the chain to screen the external spin @xmath86 . in terms of the original model , eq .
( [ h ] ) , the spins @xmath84 , @xmath85 , @xmath91 , and @xmath92 form a decoupled single - diamond state ( @xmath46 ) in the low - energy sector of the spectrum .
such local excitations are relevant relatively close to the phase boundary @xmath72 , where the coupling @xmath93 dominates the biquadratic term in eq .
( [ hii ] ) . for larger values of the parameter @xmath35 ,
the biquadratic exchange in @xmath94 becomes important .
since the energy of the quintet state @xmath95 grows up with @xmath35 ( see table 1 ) , we concentrate on the triplet state @xmath53 whose energy decreases with @xmath35 and crosses the energy level of @xmath46 at @xmath96 . as discussed in ref .
, such a local triplet state does not correspond to a stable fixed point , since the antiferromagnetic interaction of the effective spin-1 impurity with the rest of the chain is marginally relevant .
thus , one expects that the couplings to the next two spins in the chain , @xmath97 and @xmath98 , grow up to @xmath90 .
there appears another fixed point where the latter two spins also decouple from the chain in order to screen the spin of the triplet state @xmath53 . note that in sectors containing more spin-1 dimers , other types of screening are possible , as well .
for example , the spins of neighboring @xmath53 diamonds in fig .
[ td_state ] may be screened by forming a singlet state .
as before , the number of decoupled chain spins is four .
we suggest that such decoupled singlet states , composed of longer @xmath46 or @xmath53 diamond chains , control the observed instability of the sf phase upon decreasing the parameter @xmath17 .
note that similar complexes of @xmath45 diamonds can not produce the instability , since the energy of the resulting product state @xmath99[{\boldsymbol}{\sigma}_3,{\boldsymbol}{\sigma}_4]\cdots [ { \boldsymbol}{\sigma}_{l-1},{\boldsymbol}{\sigma}_l]$ ] exceeds the energy of the critical phase for arbitrary @xmath100 .
a detailed numerical study of periodic clusters containing up to @xmath78 cells close to the lines @xmath101 , @xmath102 , and @xmath103 in fig . [ ph_diagram ] suggests that for smaller values of @xmath17 the ground state belongs to the sector @xmath104 $ ] . in the special case @xmath105 ,
our numerical results reproduce the phase diagram of the frustrated sdc,@xcite where the so - called tetramer - dimer phase ( denoted as td1 in fig .
[ ph_diagram ] ) appears in the interval @xmath106 .
this doubly degenerate singlet state may be roughly represented as a product of single - diamond @xmath46 states on every second diamond , as depicted in fig .
[ td_state ] .
the short - range correlations shown in fig . [ td_state ] imply that the simple product state is a good variational wave function over the entire region occupied by the td1 phase , excluding a narrow region near the line ae .
the product state is an exact ground state at @xmath107 .
an extrapolation of the exact - diagonalization ( ed ) data suggests that the phase boundary between the td1 and sf phases lies extremely close to the @xmath102 line separating the single - diamond states @xmath45 and @xmath46 . on the other hand
, the phase boundary between the td1 and fm1 phases clearly deviates from the single - diamond boundary af .
the deviation from the line af is related to weak but finite interactions between neighboring @xmath46 diamonds appearing in the second - order perturbation theory in the parameter @xmath38 . on increasing the cyclic exchange parameter @xmath35 at fixed @xmath17 one finds another tetramer - dimer type state ( td2 ) .
the picture of the spin - spin correlations , fig .
[ td_state ] , suggests that every second diamond is approximately in the @xmath53 state .
nevertheless , the numerical analysis shows relatively strong antiferromagnetic correlations between neighboring @xmath53 diamonds , as opposed to the td1 state where the @xmath46 diamonds are weakly correlated . clearly , both tetramer - dimer phases are gapped and doubly - degenerated . according to the general rules,@xcite
one may expect a discontinuous quantum phase transition between the quantum gapped phases td1 and td2 .
numerically , the transition point is indistinguishable from the exact single - diamond phase boundary @xmath108 ( @xmath96 ) separating the single - diamond states @xmath46 and @xmath53 .
short - range spin - spin correlations vs. @xmath35 in the tetramer - dimer phases td1 and td2 for periodic chains with @xmath109 cells ( @xmath110 ) .
@xmath111 , @xmath112 , @xmath113 , @xmath114 .
, width=283 ] a numerical inspection of the short - range correlators @xmath116 in finite periodic chains implies that for moderate values of the parameter @xmath17 the established ground states belong to the sector @xmath115 $ ] .
thus , in the low - energy region the diamond model , eqs .
( [ h ] ) and ( [ h2 ] ) , is equivalent to the following mixed - spin heisenberg model ( @xmath117 , @xmath118 ) with three - spin exchange interactions [ h3 ] h_1&=&_n=1^l , where @xmath119 . to the best of our knowledge ,
generic mixed - spin quantum heisenberg models with multiple - spin exchange interactions have not been discussed in the literature , although these interactions may play an important role in some recently synthesized mixed - spin magnetic materials.@xcite here we restrict ourselves to a general overview of the spin phases in the specific case @xmath120 , and to ferromagnetic exchange interactions between the @xmath121 spins ( @xmath122 ) .
the first two terms in eq .
( [ h3 ] ) define a standard mixed - spin heisenberg model containing additional non - frustrated @xmath123 exchange bonds .
the lieb - mattis theorem@xcite predicts a ferrimagnetic ground state for this bipartite model which coincides with the classical ferrimagnetic two - sublattice nel state ( the fm2 phase in fig .
[ ph_diagram ] ) . to study the role of the competing three - spin interactions , we use a qualitative spin - wave analysis supplemented by numerical ed calculations for finite periodic chains .
spin - wave spectrum of the @xmath120 chain with three - spin exchange interactions , eq .
( [ h3 ] ) , in the ferrimagnetic ( a ) and the canted ( b ) ground states at @xmath124 and @xmath125 , respectively .
@xmath126 , @xmath127 .
, title="fig:",width=151 ] spin - wave spectrum of the @xmath120 chain with three - spin exchange interactions , eq .
( [ h3 ] ) , in the ferrimagnetic ( a ) and the canted ( b ) ground states at @xmath124 and @xmath125 , respectively .
@xmath126 , @xmath127 .
, title="fig:",width=151 ] lowest - energy excited states in a periodic chain with @xmath78 diamonds in the sectors with magnetic moments @xmath128 ( triangles ) and @xmath129 ( circles ) . @xmath130 is the magnetic moment of the ferrimagnetic ground state .
@xmath131 , @xmath126 .
, width=264 ] let us start with a discussion of the semi - classical limit of the model ( [ h3 ] ) .
since the three - spin interaction contains an additional overall factor of @xmath132 , it is convenient to redefine the coupling constant @xmath35 and measure the strength of these interactions in terms of @xmath133 .
the classical nel configuration defined by @xmath134 and @xmath135 survives up to @xmath136 . for @xmath137 , the ferromagnetic arrangement of the smaller @xmath138 spins
remains unchanged whereas the orientation of the @xmath139 spins deviates from the z axis : @xmath140 . here
@xmath141 and the azimuthal angle @xmath142 takes arbitrary values from the interval @xmath143 ( @xmath144 . as a rule , quantum fluctuations favor the planar spin configurations ( @xmath145 ; @xmath146 ) , so that one may expect a reduction of the degeneracy to @xmath147 .
such high degeneracy of the ground state is typical for a number of spin models on corner - sharing lattices . as seen in fig .
[ spectrum]b , in a spin - wave approximation the degeneracy produces a full line of zero modes @xmath148 in the brillouin zone .
note that the presence of the gapless antiferromagnetic mode @xmath149 ( @xmath150 ) for @xmath151 is related to the finite transverse magnetization of the classical canted state .
the explicit expression for @xmath152 reads _
af=2j^s_2| ( ) | , where @xmath153 .
it is clear that the antiferromagnetic mode @xmath152 is stabilized by the ferromagnetic couplings between the @xmath121 spins .
the above picture of low - lying excitations in the ferrimagnetic phase close to the phase transition point is confirmed by the numerical results at @xmath131 , fig .
[ spectrum_ed ] .
the numerical estimate for the phase transition point is @xmath154 .
short - range spin - spin correlations vs. @xmath35 for periodic chains with @xmath78 cells ( @xmath126 ) .
, title="fig:",width=132 ] short - range spin - spin correlations vs. @xmath35 for periodic chains with @xmath78 cells ( @xmath126 ) .
, title="fig:",width=132 ] additional information about the phase diagram of eq .
( [ h3 ] ) may be obtained from the behavior of the short - range correlations , fig .
[ corr ] , and the singlet - triplet excitation gap , fig .
[ gap ] , with the parameter @xmath35 . for larger @xmath155 ,
the ed results indicate at least two additional phases denoted by dm and fm3 in fig .
[ ph_diagram ] : dm is a non - magnetic gapped singlet state stabilized approximately in the region @xmath156 ( see fig . [ gap ] ) , and fm3 is a magnetic phase similar to the fm2 phase .
since the unit cell consists of three spin-1/2 variables , the lieb - schultz - mattis theorem@xcite suggests that the spin gap must be accompanied with at least doubly - degenerate ground states .
further information about the dm phase can be extracted from the dimerization operator @xmath157 where @xmath158 is the spin operator at site @xmath14 .
the lattice sites @xmath159 and @xmath160 are supposed to be symmetric under the reflection from the central site @xmath14 .
it is convenient to use the symmetric ( antisymmetric ) combinations @xmath161 , where @xmath162 is the translationally invariant singlet ground state of the finite periodic chain and @xmath163 is the singlet excited state which is almost degenerate with @xmath162 .
the states @xmath164 are not translationally invariant and may be expected to produce finite values of the dimer order - parameter @xmath165 in a dimerized system.@xcite the caricature of the dm state presented in fig .
[ dimerization ] is obtained from the extrapolation of the ed results for @xmath165 ( @xmath166 , and @xmath167 ) .
a pronounced enhancement of the dimer order - parameter with @xmath21 is indicated only for two types of bonds in the dimer model ( the thick lines in fig .
[ dimerization ] ) . clearly , the suggested ground state exhibits a four - fold degeneracy . the excitation gap in the dm phase as a function of the parameter @xmath35 . , width=264 ] finally , the numerical results point towards the existence of another partially polarized magnetic phase in the narrow interval @xmath168 between the ferrimagnetic ( fm2 ) and the dimerized ( dm ) phases ( the hatched area in fig .
[ ph_diagram ] ) . recently
, similar exotic magnetic states have been predicted in a number of one - dimensional spin systems with magnetic frustrations.@xcite typically , discussed spin states exhibit a partially polarized magnetization in the @xmath58 direction ( @xmath169 ) , a quasi - long - range transverse magnetic order , and a gapless linear mode related to the destroyed classical canted state .
unfortunately , the methods used in the present study do not suggest a clear statement indicating the existence of such an exotic phase in the discussed system .
caricature of the dimerized singlet phase dm as obtained from an extrapolation of the numerical ed results for periodic chains with @xmath166 , and @xmath167 .
the open circles ( thick lines ) denote the symmetric ( antisymmetric ) eigenstates of two spin-1/2 operators .
@xmath170 , @xmath171 , @xmath120 . ,
in conclusion , we have examined the impact of the cyclic four - spin exchange interactions on the ground - state phase diagram of the symmetric spin-1/2 diamond chain . using the local symmetries of the model
, the spin phases were classified by the set of good quantum numbers @xmath22 ( @xmath23 ) related to the composite spins @xmath24 .
the presented study demonstrates a rich phase diagram in the parameter space @xmath172 .
apart from the standard magnetic and paramagnetic phases , the system exhibits two different tetramer - dimer phases in the sector @xmath173 $ ] as well as an exotic four - fold - degenerate dimerized ground state in the sector @xmath174 $ ] .
this research was supported by deutsche forschungsgemeinschaft ( grant 436 bul 17/9/06 ) and the bulgarian science foundation ( grants f1414 and d002 - 264/18.12.08 ) .
part of the work has been done in max - planck - institut fr physik komplexer systeme , dresden .
s. notbohm , p. ribeiro , b. lake , d. a. tennant , k. p. schmidt , g. s. uhrig , c. hess , r. klingeler , g. behr , b. bchner , m. reehuis , r. i. bewley , c. d. frost , p. manuel , and r. s. eccleston , phys .
lett . * 98 * , 027403 ( 2007 ) .
t. hikihara , t. momoi , and x. hu , phys .
* 90 * , 087204 ( 2003 ) ; v. gritsev , b. normand , and d. baeriswyl , phys .
b * 69 * , 094431 ( 2004 ) ; v. n. kotov , d. x. yao , a. h. castro neto , and d. k. campbell , arxiv:0704.0114v1 ( unpublished ) ; a. w. sandvik , phys .
98 * , 227202 ( 2007 ) .
k. takano , k. kubo , and h. sakamoto , j. phys . : condens
. matter , * 8 * , 6405 ( 1996 ) ; h. niggemann , g. uimin , and j. zittartz , ibid .
, * 9 * , 9031 ( 1997 ) ; a. honecker and a. luchli , phys .
b * 63 * , 174407 ( 2001 ) . n. b. ivanov and j. richter , phys
a * 232 * , 308 ( 1997 ) ; j. richter , n. b. ivanov and j. schulenburg , j. phys . : condens .
matter * 10 * , 3635 ( 1998 ) ; a. koga , k. okunishi , and n. kawakami , phys . rev .
b * 62 * , 5558 ( 2000 ) .
interestingly , the multiple - spin exchange term in eq .
( [ h3 ] ) also appears in some integrable alternating quantum spin @xmath175 chains : see , e.g. , g. a. p. ribeiro and a. klmper , nuclear physics b * 801 * [ fs ] , 247 ( 2008 ) , and references therein . n. b. ivanov and j. richter , phys .
b * 69 * , 214420 ( 2004 ) ; s. yoshikawa and s. miyashita , j. phys .
jpn . * 74 * suppl . , 71 ( 2005 ) ; k. hida , j. phys . sos . jpn .
* 76 * , 024714 ( 2007 ) ; k. hida and k. takano , phys .
b * 78 * , 064407 ( 2008 ) . | we study the phase diagram of a symmetric spin-@xmath0 heisenberg diamond chain with additional cyclic four - spin exchange interactions .
the presented analysis supplemented by numerical exact - diagonalization results for finite periodic clusters implies a rich phase diagram containing , apart from standard magnetic and spin - liquid phases , two different tetramer - dimer phases as well as an exotic four - fold degenerate dimerized phase .
the characteristics of the established spin phases as well as the nature of quantum phase transitions are discussed , as well . |
observations of quasars above 2 kev have in the past been largely devoted to objects selected via x - rays .
their 210 kev spectral indices cluster tightly around a `` canonical '' value of 0.9 ( @xmath8 ; williams et al .
1992 , comastri et al .
1992 , reeves et al .
1997 , lawson & turner 1997 ) .
this uniformity may be the result of a selection effect ( elvis 1992 ) .
seyfert galaxies , in fact , do indeed show a significantly wider range of 210 kev spectral indices ( brandt , mathur & elvis 1997 ) and qsos might show the same effect as reported by recent studies of asca observations ( george et al .
we have therefore started a program to observe , with bepposax ( boella et al .
1997a ) , a reasonably large , well defined , and representative sample of optically selected quasars .
our goals are to perform a systematic study of their hard ( @xmath9 kev ) emission spectra and , taking advantage of the broad bepposax band , to compare them with the soft 0.12 kev spectra acquired simultaneously .
bepposax has so far observed 10 quasars extracted from the laor et al .
( 1997 ) sample , nine as part of a core program and one as part of the science verification phase ( svp ) .
the laor et al .
sample consists of 23 pg quasars selected to have @xmath10 , galactic @xmath11 @xmath12 and @xmath13 .
these quasars all have good rosat pspc x - ray spectra , radio fluxes , ir photometry , high s / n optical spectro photometry , iue and hst spectra ( laor et al . 1997 and
references therein ) .
the 0.22 kev pspc spectra of most of these 23 quasars are well fitted by a single power law plus galactic absorption .
the average spectral index is @xmath14 , and the @xmath15 range is 0.92.8 .
the large spread in @xmath15 permitted the discovery of correlations between the soft x - ray spectral shape and optical emission line properties ( laor et al .
1994 , 1997 , boller , brandt & fink 1996 , ulrich - demoulin & molendi 1996 ) : steeper soft x - ray quasars tend to have narrower @xmath16 lines , fainter [ oiii ] emission and stronger feii emission with respect to @xmath16 .
laor et al .
( 1997 ) interpret the @xmath17 fwhm anticorrelation in terms of a dependence of @xmath15 on @xmath18 .
the line width is inversely proportional to @xmath19 if the broad line region is virialized and if its size is determined by the central source luminosity ( see laor et al .
1997 , 4.7 ) .
so narrow - line , steep ( 0.12 kev ) spectrum agns emit close to the eddington luminosity ( nicastro 2000 ) . a proposed scenario for these sources , as described by pounds et al .
( 1995 ) , is that the hard x - ray power law is produced by comptonization in a hot corona : as the object becomes more luminous in the optical - uv , compton cooling of the corona increases .
the corona becomes colder , thus producing a steeper x - ray power law .
if this mechanism is operating , then steep @xmath15(pspc ) quasars should also have a steep hard x - ray power law that would have been missed by hard x - ray surveys .
bepposax 0.110 kev observations should help in verifying this picture ; test if the correlations between the soft x - ray spectrum and the optical lines properties hold also at high x - ray energies ; and study the iron k emission lines in high luminosity objects , to test the claim of an x - ray baldwin effect ( iwasawa & taniguchi 1993 , nandra et al .
1997 ) .
in this paper we report the analysis of bepposax observations of 10 pg quasars . in
2 we describe the observations and the data analysis ; in 3 we present the broad band 0.1100 kev spectra ; in 4 we discuss in detail the 0.110 kev spectral shape and compare it with the pspc results ; in 5 we present results on the iron line ; in 6 we present our conclusions .
a correlation between the 0.110 kev spectral shape and the optical line properties is deferred to a future publication . in the paper @xmath20=50 km
s@xmath21 mpc@xmath21 and @xmath22=0.5 are assumed .
table [ tab1 ] lists the pg quasars observed by bepposax , together with their redshifts ( schmidt & green 1983 ) and 21 cm galactic column densities ( we use the values quoted in laor et al .
1997 , where the relevant references are also reported ) .
pg 1226@xmath6023 ( 3c 273 ) was observed during the svp , and a detailed wide band spectral analysis has been published by grandi et al .
( 1997 ) .
.pg quasars observed by bepposax [ cols="<,<,^,^ " , ] @xmath23 fit 1 : power law plus galactic absorption ; fit 2 : same as fit 1 plus a narrow line with energy relative to the quasar frame .
we presented the spectral analysis of the bepposax observations of 10 pg quasars selected from the laor et al .
( 1997 ) sample .
the main results can be summarized as follows : * together with pg 1226@xmath6023 a positive detection of the continuum in the 15100 kev energy range has been found in two more quasars : pg 0947@xmath6396 and pg 1115@xmath6407 . however , a possible contamination from hard serendipitous sources in the pds field of view can not be ruled out with the present data . *
the distribution of the 210 kev power law energy indices is similar to that observed in other quasars samples .
the dispersion around the average value of @xmath24 reflects the large spread of the best fit values over the range 0.5 @xmath25 1.5 . *
no intrinsic absorption has been detected in any of the objects since the absorbing columns are always compatible with the galactic values . *
significant spectral curvature is present in the bepposax spectra of most of the quasars and is related to the statistics of the observed spectrum : sources with the same energy index over the 0.110 kev band are fainter .
the average curvature can be parameterized as a spectral flattening by @xmath26 towards high energies requiring two component models .
the exact spectral shape and intensity of these components vary from object to object . in a few cases the curvature is due to a strong soft excess `` below about 1 kev , while for pg 1626@xmath6554 , independent lecs
pspc fits suggest the presence of an ' ' ultra soft excess " ( below 0.3 kev ) .
the curvature in pg 1402@xmath626 is mainly due to a hard tail above @xmath2 5 kev rather than a soft excess " ; also pg 1626@xmath6554 shows evidence for a hard tail
. a spectral hardening at even higher energies ( @xmath7 10 kev ) could be present in the two quasars detected in the pds band .
* the origin of the spectral curvature is likely due to the combined effect of thermal emission from an accretion disk peaking in the far uv and the onset of a compton reflection component at high energies .
the addition of these components does provide a better description of the observed spectra .
we also note that quasars with a steeper 0.12 kev spectrum tend to be steeper also in the 210 kev band , although the effect is detected only at the 68 % confidence level . *
iron k@xmath5 lines are detected in 4 quasars .
the rest frame line energy ( 6.7 kev ) and equivalent width ( 580 ev ) of pg 1115@xmath6407 are consistent with those found in a few low redshift narrow line seyfert 1 galaxies ( comastri et al .
1998 , turner et al .
1998 , vaughan et al .
1999 , leighly 1999 ) .
the contemporaneous presence of a steep 0.110 kev continuum and narrow h@xmath27 line allows us to classify this object as a relatively high redshift nlsy1 .
+ for the optically broad lined quasar pg 0947@xmath6396 , the rest frame line energy of 6.4 kev is similar to that found in many local seyfert 1 galaxies , suggesting fluorescence from cold iron .
the line equivalent width ( 670 ev ) is however higher ( at the @xmath28 level ) than that usually found in seyfert 1 galaxies .
we investigate the possibility that the high ew of the line , as well as an excess of counts detected in the pds , are the result of fitting a complex continuum with a simple power - law model .
we tried several other models ( the inclusion of a compton reflection component or of a thick and partial covering absorber ) but in all cases the line is still required and its ew is never smaller than 400 ev . + a line has also been detected in pg 1352@xmath6183 and pg 1226@xmath6023 . in the first case the line energy and intensity are compatible with the values detected in many seyfert galaxies . in the second case it could be witnessing a seyfert - like spectrum diluted in the jet emission ( grandi et al .
1997 , haardt et al . 1998 ) .
+ the detection of significant fe k shell emission in 3 radio quiet quasars with 210 kev luminosities in the range 15 @xmath29 10@xmath30 erg s@xmath21 seems to be inconsistent with the trend seen in other radio quiet agn ( iwasawa & taniguchi 1993 ; nandra et al .
1997 , george et al .
2000 ) , where iron lines are more frequently found in lower luminosity objects .
* a detailed comparison with the recent asca observations of a larger sample of radio quiet quasars ( george et al . 2000 ) is not possible owing to the different responses , sensitivities and energy ranges covered by asca and bepposax .
we note however that there is a good agreement between the two samples especially for what concerns the average 210 kev slope and intrinsic dispersion and the presence of curved convex spectra in most of the objects .
the detection with bepposax of a strong iron line in a few relatively high luminosity objects deserves further investigation .
the foreseen xmm - newton quasar surveys will most likely settle several open issues .
tm and ff thank a. wolter for the useful discussion on the lecs
pspc intercalibration and p.ciliegi for the several comments and suggestions to the paper .
wnb acknowledges the support of nasa ltsa grant nag5 - 8107 , and ac acknowledges the support of italian space agency contract asi - ars-98 - 119 and murst grant cofin-98 - 02 - 32 .
the discrepancy between the bepposax soft spectral index and the pspc one can be due to systematic errors in the instrument intercalibration . to investigate this discrepancy we fitted the lecs spectra in the energy range 0.11.5 kev with a power law plus galactic absorption . considering the different response curves of the two instruments around 1 kev ( the lecs effective area increases from 1 to 2 kev , the pspc area decreases in the same range ) this is the lecs range which is more appropriate for a quantitative comparison with the pspc . for the pg 1226@xmath6023 and pg 1202@xmath6281 fits we have added an edge at 0.6 kev and 0.8 kev respectively in the model .
the resulting energy indices ( @xmath31 ) are shown in fig .
[ fig3 ] as a function of the pspc ones , together with the @xmath32(pspc ) line . to increase the sample we have added four bl lacertae objects observed by bepposax ( wolter et al .
bl lacs objects usually do not show large neutral or ionized low energy absorption ( unlike seyfert 1 and 2 galaxies ) the four bl lacs were selected from the ten of the wolter et al .
sample , excluding the objects with a complex ( or variable ) soft x - ray spectrum .
a systematic difference between the lecs and pspc spectral indices is again evident .
the bepposax soft spectral indices vs the pspc ones can be well fitted by a line with slope _
[email protected] and offset @[email protected] . fixing _
m _ to 1 gives a constant difference between the two detectors measurements of @xmath34 .
excluding the most discrepant point makes a marginal difference .
this discrepancy is smaller than that found between the pspc and the einstein ipc ( @xmath35=0.5 ; fiore et al .
1994 ) and the one between the pspc and the asca sis ( @xmath36 ; iwasawa et al .
1999 ) . in these latter two cases
part of the difference can be due to the fact that the common energy range between these instruments is not large .
conversely , the energy range shared by the pspc and the lecs is wide ( 0.12 kev ) , and , in particular , both instruments have good sensitivity below the carbon edge at 0.28 kev .
their comparison is therefore more straightforward .
however , if quasar spectra deviate from the simple power law , a small difference in the spectral index measured by the two instruments , that look at different average energies , could be possible . in order to check this effect
we have fitted the lecs and pspc spectra of the quasars with the best photon statistics simultaneously with curved models ( broken power law and the power law with energy index variable with energy presented in laor et al 1994 ) , allowing for a different normalization in the two instruments and letting the model parameters linked together .
the discrepancy in the estimated spectral index of about 0.2 could result from the differences in instrument response functions coupled with an energy dependent spectral slope at 0.12 kev ( with a reasonable curvature ) in about half of the cases .
it is unclear then that the pspc and bepposax results are always inconsistent .
figure [ fig4 ] shows the bepposax 2 kev luminosity as a function of the pspc 2 kev luminosity .
six quasars have bepposax luminosities slightly higher than pspc luminosity : the bepposax average luminosity is 30 % higher than the pspc one .
variability can certainly play a role in this difference , given the small sample of objects .
however we note that 2 kev are the limit of the pspc band and therefore a small uncertainty on the pspc spectral shape could translate in a relatively large error on the 2 kev flux and luminosity . the slope of the correlation in fig .
[ fig4 ] is consistent with 1 , indicating that this possible shift should not alter the results of the correlations between luminosity and other quasar properties discussed in laor et al .
( 1997 ) .
boella g. , butler r.c .
, perola g.c . , et al .
, 1997a , a&as 122 , 299 boella g. , chiappetti l. , conti g. , et al . , 1997b , a&as 122 , 327 boller t. , brandt w.n . ,
fink h.h . , 1996 ,
a&a 305 , 53 brandt w.n . ,
laor a. , wills b.j . , 2000 , astro - ph/9908016 brandt w.n . , mathur s. , elvis m. , 1997 , mnras 285 , l25 chiappetti l. , cusumano g. , del sordo s. , et al .
, 1998 , in `` the active x - ray sky '' , l. scarsi , h. bradt , p. giommi , f. fiore ( eds . ) , nuclear physics b proc .
69/1 - 3 , 610 ciliegi p. , maccacaro t. , 1997 , mnras 292 , 338 ciliegi p. , maccacaro t. , 1996 , mnras 282 , 477 comastri a. , fiore f. , guainazzi m. , et al . , 1998 , a&a 333 , 31 comastri a. , setti g. , zamorani g. , et al . , 1992 ,
apj 384 , 62 elvis m. , fiore f. , siemiginowska a. , bechtold j. , mathur s. mcdowell j.c.m . , 2000 , apj submitted elvis m. , 1992 , in frontiers of x - ray astronomy , univ acad . press inc .
[ tokyo ] ed .
y. tanaka & k. koyama , 567 ebeling h. , voges w. , bohoringer h. , et al . 1996 , mnras 281 , 799 fiore f. , la franca f. , giommi p. , et al . , 1999a , mnras 306 , l55 fiore f. , guainazzi m. , grandi p. , 1999b
, + ftp://www.sdc.asi.it/pub/sax/doc/software_docs/saxabc_v1.2.ps.gz fiore f. , mineo t. , laor a. , giallongo e. , 1998 , in `` the active x - ray sky '' , l. scarsi , h. bradt , p. giommi , f. fiore ( eds . ) , nuclear physics b proc . suppl .
69/1 - 3 , 529 fiore f. , elvis m. , mcdowell c. , siemiginowska a. , wilkes b. , 1994 , apj 431 , 515 frontera f. , costa e. , dal fiume d. , et al . , 1997 , a&as 122 , 357 george i.m . , turner t.j .
, yaqoob t. , et al . ,
2000 apj in press astro - ph/9910218 grandi p. , guainazzi m. , mineo t. , et al . ,
1997 , a&a 325 , l17 haardt f. , fossati g. , grandi p. , et al . , 1998 ,
a&a 340 , 35 iwasawa k. , fabian a.c . , nandra k. , 1999 , mnras 307 , 611 iwasawa k. , taniguchi y. , 1993 , apj 370 , l61 kellermann k.i .
, saramek r. , schmidt m. , shaffer d.b .
, green r. , 1989 , aj 98 , 1195 laor a. , fiore f. , elvis m. , et al . , 1997 , apj 477 , 93 laor a. , fiore f. , elvis m. , et al .
, 1994 , apj 435 , 611 lawson a.j .
, turner m.j.l . , 1997 ,
mnras 288 , 920 leighly k.m . , 1999
apjs 125 317 levine a.m. , lang f.l .
, lewin w.h.g . , et al . , 1984 , apjs 54 , 581 maccacaro t. , gioia i. m. , wolter a. , zamorani g. , stocke j. t. , 1988 , apj 326 , 680 manzo g. , giarrusso s. , santangelo a. , et al . , 1997 , a&as 122 , 341 morrison r. , mccammon d. , 1983 , apj 270 , 119 nandra k. , george i.m .
, mushotzky r.f . , et al . , 1997 , apj 488 , 91 nicastro f. , 2000 , apj in press astro - ph/9912524 orr a. , yaqoob t. , parmar a.n . , et al . , 1998 , a&a 337 , 685 parmar a.n . , oosterbroek t. , orr a. , et al . , 1998 , a&as 136 , 407 parmar a.n . ,
martin d.d.e . , bavdaz m. , et al . , 1997 , a&as 122 , 309 pounds k.a .
, done c. , osborne j.p . , 1995 ,
mnras 277 , l5 reeves j.n .
, turner m.j.l . , ohashi t. , kii t. , 1997 , mnras 292 , 468 schmidt m. , green r.f .
, 1983 , apj 413 , 116 stocke j.t . , morris s.l . ,
gioia i.m . , et al . , 1991 , apjs 76 , 813 stocke j.t . , liebert j. , schmidt g. , et al . , 1985 , apj 289 , 619 turner t.j . , george i.m . , nandra p. , 1998
, apj 508 , 648 ulrich - demoulin m.h .
, molendi s. , 1996 , apj 457 , 77 vaughan s. , reeves j. , warwick r. , edelson r. , 1999 , mnras 309 , 113 williams o.r .
, turner m.j.l . ,
stewart g.c .
, et al . 1992 , apj 389 , 157 wolter a. , comastri a. , ghisellini g. , et al . , 1998 , a&a 335 , 899 yuan w. , brinkmann w. , siebert j. , voges w. , 1998 , a&a 330 , 108 | we present results from bepposax observations of 10 low redshift quasars .
the quasars are part of the laor et al .
( 1997 ) sample of 23 optically selected pg quasars with redshift @xmath0 and low galactic absorption along the line of sight . significant spectral curvature is detected for the 6 quasars with the highest signal to noise ratio in their low energy spectra .
the average curvature can be parameterized as a flattening of the underlying power law by @xmath1 above @xmath2 1 kev .
we find that quasars with a steeper soft x - ray ( 0.12 kev ) spectrum tend to be steeper also at higher ( 210 kev ) energies .
the distribution of the best fit 210 kev slopes is similar to that found for other radio - quiet agn and characterized by a large dispersion around the mean : @xmath3 1.0 @xmath4 0.3 .
iron k@xmath5 lines are detected in 4 quasars . in the narrow line quasar pg 1115@xmath6407
, the rest frame line energy ( [email protected] kev ) and equivalent width ( 580@xmath4280 ev ) are consistent with those found in a few low redshift narrow line seyfert 1 galaxies ( nlsy1 ) .
this , together with the similarity of the 0.110 kev x - ray continuum , suggests that this quasar is the higher redshift and luminosity analog of a nlsy1 . in the broad line quasar
pg 0947@xmath6396 , the rest - frame line energy suggests fluorescence from cold iron . the line equivalent width ( @xmath7 400 ev ) is however about 23 times higher than that usually found in seyfert 1 galaxies . the high energy power - law slopes and
the iron line emission properties seem to be unrelated to the x ray luminosity . |
the proposed laser interferometer space antenna ( lisa ) @xcite , is expected to provide a variety of high - precision gravitational wave ( gw ) measurements .
one of the most interesting targets for this space - based detector are the gws generated by stellar - mass compact objects inspiralling into ( super)massive black holes [ ( s)mbhs to ( a few)@xmath1 .
we emphasize this point because the name `` supermassive black hole '' is often taken to refer to a hole with mass in the range @xmath2 .
gw events from such black holes will typically be at frequencies too low for lisa s sensitivity . ] ] .
accumulated astrometric observations provide strong support in favor of the existence of a `` dark '' compact object in the core of every galaxy ( for which the central parsec region can be resolved ) @xcite . with masses ranging between @xmath3 these objects
are believed to be massive kerr black holes @xcite .
it is also believed that these holes are the `` quiet '' remnants of an older quasar population @xcite .
multi - body interactions in the `` cusp '' stellar population surrounding these smbhs can put stellar mass compact objects onto orbits that come close to the central black hole .
if the object passes sufficiently close to the ( s)mbh , it may be captured and subsequently spiral in by the emission of gws @xcite .
initially , the captured bodies are expected to be on `` generic '' orbits , i.e. , eccentric ( with eccentricity @xmath4 ) and inclined with respect to the central black hole s equatorial plane @xcite .
these orbits evolve adiabatically due to gw emission , decreasing in eccentricity and periastron while the inclination of the orbit remains approximately constant , but increases slightly ( see @xcite for an approximate description of the full inspiral ) . for central black hole masses in the range @xmath5 , the gws emitted during the inspiral will be at frequencies close to the floor of the lisa noise curve ( @xmath6 mhz ) .
lisa will detect the bursts of radiation emitted near periapse throughout the inspiral , but these will not be individually resolvable @xcite unless they occur in our galaxy @xcite . however , during the last few years of inspiral , when the small object is orbiting deep inside the smbh s gravitational field , gws are radiated continuously in the lisa band . during this phase ,
the residual eccentricity is typically at @xmath7 @xcite and the orbital motion will exhibit extreme versions of relativistic effects ( e.g. , periastron and lense - thirring precession ) . as a consequence
, the resulting gw signal will be strongly `` colored '' by these effects and will take a complicated form @xcite . by complicating the waveform , these strong - field effects potentially make data analysis difficult ; but , they also encode a great deal of information about the strong field nature of the spacetime . by accurately measuring all of these effects , it is expected that we will be able to `` map '' the spacetime of the black hole @xcite , probing its multipolar structure and verifying that it obeys the constraints that general relativity imposes on black hole solutions @xcite .
the expected amplitude of the signal from these extreme - mass - ratio inspirals ( emris ) is about an order of magnitude below lisa s projected instrumental noise and , at low frequencies ( @xmath8 mhz ) , is several orders of magnitude below `` confusion noise '' produced by unresolved galactic binaries @xcite .
however , the signals will be observed for @xmath9 waveform cycles , and matched filtering will therefore allow detection of these signals with high signal - to - noise ratio ( snr ) out to a redshift @xmath10 @xcite .
preliminary estimates suggest lisa could see as many as @xmath11 emri events during its lifetime @xcite , using a suitable `` semi - coherent '' search algorithm and provided that confusion noise can be efficiently reduced in the real data .
matched filtering algorithms require the correlation of the detector s output data stream with a bank of waveform templates which describe the real signal with sufficient accuracy , covering the whole parameter space .
the fact that we are dealing with a binary system of extreme mass ratio @xmath12 means that the gravitational waveform may be obtained accurately using black hole perturbation theory .
the extreme mass ratio also guarantees that the orbital parameters evolve adiabatically under radiation reaction ; i.e. , they evolve on a much longer time scale than the orbital periods .
this implies that within the radiation reaction time scale , the inspiral waveform can be approximated by `` snapshot '' waveforms waveforms calculated by assuming that the small object is moving along a geodesic , neglecting backreaction for that short stretch of time .
these snapshots are constructed using the teukolsky equation @xcite , an equation that describes the first order change to the curvature tensor of a black hole due to some perturbing source .
accurate teukolsky - based ( tb ) snapshot waveforms have been calculated for inclined - circular orbits @xcite , equatorial - eccentric orbits @xcite and most recently , for a certain number of generic ( inclined - eccentric ) orbits @xcite .
the reader can find recent reviews on the modeling of emri waveforms and orbital dynamics in refs .
@xcite .
tb waveforms are computationally expensive to generate as they require the numerical integration of the teukolsky equation and summation over a large number of multipole modes . in this sense , these waveforms are not `` user - friendly '' , especially when it comes to realistic data analysis calculations where one has to handle a bank which contains @xmath13 of these waveforms @xcite .
moreover , the teukolsky formalism does not provide any information on `` conservative '' self - interaction effects . to compute these
, one must use a self - force formalism .
this approach is still under development and is very likely to be even more computationally expensive when it is completed ( see @xcite for a recent review ) .
these difficulties have led to a need for the construction of approximate families of waveforms that capture the main features of the true signals , but which can be generated quickly in large numbers .
such approximate waveforms are already being used for scoping out data analysis algorithms for the detection of emris in lisa data @xcite , and may ultimately play some role as fiducial detection templates in the final data analysis .
one possible approach is to construct post - newtonian ( pn ) waveforms , which have the advantage of being analytic and therefore very easy to generate .
post - newtonian emri waveforms have been computed in the schwarzschild spacetime for both quasi - circular @xcite and eccentric orbits @xcite , and in the spacetime outside a slowly rotating kerr black hole for quasi - circular orbits @xcite .
recently another class of approximate waveforms have become available , based on various `` kludge '' approaches @xcite .
the basic idea of the kludges is to combine different prescriptions for the orbital evolution and waveform emission ( not necessarily in a self - consistent way ) .
an `` analytic kludge '' ( ak ) was developed by barack and cutler @xcite ( see also @xcite ) . in their model
, the small object is moving on a keplerian orbit , amended to include the effects of pericenter precession , lense - thirring precession , and inspiral from radiation reaction .
the emitted gravitational waveforms are described by the lowest - order quadrupole formula . in this paper
, we consider an alternative way to construct kludge waveforms .
this approach is much less amenable to analytic calculation , so we called it the `` numerical kludge '' ( nk ) .
the first step of the nk is to produce an inspiral trajectory in `` phase space'' the space defined by the constants ( orbital energy , axial angular momentum , and `` carter constant '' ) which specify orbits ( up to initial conditions ) @xcite .
the second step is to numerically integrate the kerr geodesic equations along this inspiral trajectory and hence obtain the boyer - lindquist coordinates of the inspiralling object as a function of time @xcite .
the final step is to construct a gravitational waveform from this inspiral trajectory .
the approach to waveform construction that we take is to identify the boyer - lindquist coordinates of the source with spherical polar coordinates in flat - space .
there are several different expressions available in the literature for the gravitational waveforms from orbits in flat space , and we apply these to our pseudo - flat - space trajectory to construct waveforms . specifically , we look at the quadrupole formula @xcite , which is valid in the limit of weak - field ( i.e. , flat - space ) and slow motion .
we also examine the quadrupole - octupole formula of bekenstein @xcite , as well as a formula derived by press @xcite .
the press formula is also a weak - field expression , but is not restricted to slow - motion or small sources , and contains radiation at orders higher than quadrupole and octupole .
the purpose of this paper to establish and delimit the accuracy and reliability of waveforms constructed in these various ways .
we do this by comparing to tb waveforms . tb waveforms are currently the most accurate emri waveforms available . in most cases ,
tb waveforms represent the emission from geodesic orbits we mostly do not include the radiative evolution of orbital parameters in this work .
( with one exception : because complete tb inspirals exist for zero - eccentricity orbits @xcite , we compare to a full inspiral in this case . )
we compare the various nk waveforms with tb waveforms using an overlap integral which weights the waveforms in frequency space by the expected lisa noise curve , and maximizes the overlap with respect to time offsets .
this overlap is identical to the test used to evaluate the efficiency of model waveforms as detection templates .
we find that quadrupole - octupole nk waveforms are in remarkable agreement ( overlaps@xmath14 ) with tb waveforms for orbits with periastron @xmath15 .
most orbits in the final year of the inspiral satisfy this restriction which means that nk waveforms are quite accurate over a considerable portion of the inspiral parameter space . for orbits that come below that radius ( mostly prograde orbits around rapidly spinning holes ) ,
the agreement rapidly degrades ( although nk waveforms remain more accurate than post - newtonian or ak waveforms ) .
this is not surprising for such orbits the full tb waveform receives significant contributions from higher multipoles and back - scattering from the spacetime .
these effects are _ ab initio _ absent in the nk prescription .
waveforms generated using the press formula perform similarly well for @xmath15 , and slightly better than the quadrupole - octupole prescription for orbits where the contribution from higher harmonics is not negligible .
the improvement in this relatively strong - field regime is only slight , however we did not find any regime where the press formula is a significant improvement on the quadrupole - octupole prescription .
we conclude that nk waveforms can accurately reproduce true gravitational waveforms in a large part of parameter space ; but , there is little gain from going beyond the quadrupole - octupole prescription .
we also calculate kludge radiation _ fluxes _ by combining the quadrupole energy flux formula with exact geodesic motion and averaging in time ; such an analysis was done for schwarzschild orbits in @xcite .
the resulting fluxes typically compare well to the best available pn formula in the weak field , but provide a somewhat better estimate of the energy and angular momentum fluxes ( comparing to tb results ) for strong - field orbits .
these fluxes also allows us to assess by how much the approximations which go into the kludge construction are inconsistent .
since the inspiral trajectory and waveform construction are considered separately , kludge gws carry a different amount of energy and angular momentum to infinity than the inspiralling particle loses ( which is set by the formulas which determine the inspiral through orbital parameter space ) .
it is important to bear this in mind when using kludge waveforms in computations , e.g. , for estimating snrs of lisa emri detections , and to have an estimate of the size of the inconsistency .
finally , we describe how the kludge waveforms may be improved by including some of the conservative self - force effects , which may have a significant imprint on true inspiral waveforms .
this paper is organised as follows .
section [ inventory ] provides a review of existing emri waveforms , paying special attention to the kludge semi - relativistic waveforms .
details of how these waveforms can be generated are described . in section
[ overlaps ] we discuss the overlap function between two waveforms , as this will be our main tool for comparison .
section [ results ] contains the results from the comparison between kludge and teukolsky - based waveforms and fluxes . in section [ cons_pieces ] a method is outlined for including the conservative parts of the self - force in our kludge scheme .
finally , we present a concluding discussion in section [ conclusions ] . we shall use greek letters ( @xmath16 , @xmath17 , etc . ) to denote spacetime indices , and latin letters ( @xmath18 , @xmath19 , etc . ) to denote spatial indices ( unless explicitly stated otherwise ) . throughout the paper we adopt geometric units @xmath20 .
presently , several types of emri waveforms are available . in broad terms , these waveforms fall into three categories : ( i ) those calculated numerically within the framework of black hole perturbation theory ( teukolsky - based , or tb waveforms ) , ( ii ) analytic waveforms which result from self - consistent pn expansions of the gw equations _ and _ orbital motion , and ( iii ) approximate semi - relativistic waveforms , or `` kludges '' .
this third category is the focus of this paper .
kludge waveforms are constructed by combining a flat spacetime wave - emission formula with either a fully relativistic treatment of particle motion ( resulting in the numerical kludge , or nk , waveforms ) , or some analytic approximation of this motion ( leading to the analytic kludge , or ak , waveforms ) .
we shall examine the construction of nk waveforms , comparing them to tb waveforms ( the most accurate emri waveforms presently available ) . for completeness and background to this paper s discussions , we now briefly discuss each of the above waveform families . the primary framework for black hole perturbation theory in a kerr background is the teukolsky formalism @xcite , which encapsulates all gravitational radiative degrees of freedom in a single `` master '' wave equation the `` teukolsky equation ''
for the weyl scalars @xmath21 and @xmath22 .
a key feature of this equation is that it admits separation of variables in the frequency domain , which effectively reduces it to a pair of ordinary differential equations .
there are extensive results in the literature on solutions of the teukolsky equation , starting from press and teukolsky @xcite , detweiler @xcite and sasaki and nakamura @xcite .
more recent work uses the teukolsky - sasaki - nakamura formalism , see refs .
@xcite for detailed discussions and references on the subject . to date
, the teukolsky equation has been solved in the frequency domain for circular - inclined orbits @xcite , eccentric - equatorial orbits @xcite and most recently for a handful of generic ( eccentric - inclined ) orbits @xcite .
we make use of the waveforms generated by these various authors in this paper to assess the quality of our nk waveforms .
recently , the teukolsky equation has also been solved directly in the time domain @xcite .
time - domain calculations have the great advantage of speed , since they avoid the need for fourier decomposition and summation over all orbital frequency harmonics . however , these calculations are not yet mature enough to provide accurate waveforms from kerr orbits , because of the difficulty of representing the various @xmath23-functions appearing in the energy - momentum tensor of a point particle . as a final remark , we should mention again that in all the above tb calculations ( either in the frequency or the time domain ) the motion of the small object is taken to be strictly geodesic .
this is a reasonable first approximation since for an emri system the orbital evolution is adiabatic , i.e. , over a time interval of several orbits the motion is geodesic to high precision .
however , for longer stretches of time ( @xmath24 ) the effects of gravitational back - reaction become significant and can not be ignored anymore .
waveforms that take into account an evolving orbit ( and the conservative self - interaction ) require the computation of the gravitational self - force ( see @xcite for an up - to - date review and a full list of references )
. however , self - force waveforms are not yet available and are unlikely to be for the next few years .
moreover , it is very likely that self - force calculations will remain quite computationally intensive ; as such , it is very likely that it will not be possible to generate self - force based waveforms in sufficient numbers to be used for lisa data analysis .
it is therefore essential to investigate approximate , easy - to - use waveform models .
most available analytic waveforms are based on post - newtonian expansions of the orbital dynamics and wave emission , an expansion that is of greatest relevance when the bodies are widely separated .
these waveforms are typically constructed for a specific object or for restricted orbital motion .
a significant amount of work has been done on modeling waveforms from two spinning bodies with comparable masses orbiting in precessing quasi - circular orbits ( a key gw source for both ground- and space - based detectors ) .
kidder @xcite has investigated the effects of spin - spin and spin - orbital coupling on the waveform from inspiralling binaries .
the most recent investigation of pn waveforms ( and their application to data analysis ) for spinning binary systems may be found in refs .
@xcite ( see also references therein ) .
another promising approach is described in @xcite , which is the first attempt to extend the `` effective - one - body '' method @xcite to spinning compact objects . in parallel to modeling spinning binaries in quasi - circular orbits ,
there has been significant progress in the construction of waveforms for eccentric , comparable mass binaries @xcite . presently , there are no post - newtonian waveforms which include both eccentricity of the orbit and spins of the orbiting bodies .
post - newtonian models are ultimately not so useful for modeling emris , since most of the gws observable to lisa are generated from a strong field region ( @xmath25 a few @xmath26 ) , where the pn expansion is unlikely to be reliable .
one can , however , construct pn waveforms in the emri limit , accepting their certain unreliability as a way to develop a very `` quick and dirty '' set of tools for studying these waves .
pn emri waveforms are available for systems of a test mass in a quasi - circular @xcite or eccentric @xcite orbit around a schwarzschild black hole , or in a quasi - circular orbit around a slowly rotating kerr black hole @xcite .
more recently , a class of approximate pn waveforms has been developed by barack and cutler @xcite .
these `` analytic kludge '' ( ak ) waveforms are essentially phenomenological waveforms they are constructed using the classic quadrupole waveforms for eccentric keplerian orbits derived in @xcite , but with the relativistic effects of pericenter precession , lense - thirring precession , and inspiral imposed . though not as accurate as the nk waveforms described in this paper , the ak are very quick to generate , and have proven to be useful for computing the fisher information matrix in investigations of parameter measurement with emri gws @xcite .
the overlap between ak and nk waveforms is high in the very weak field , but degrades as the orbital periapse is decreased ( see @xcite ; more detailed comparisons will be included in future papers on the semi - coherent algorithm currently in preparation ) . even for geodesic orbits , ak and nk waveforms with the _ same physical parameters _ drift out of phase quickly , since the frequency structure of the two waveform families is different .
this arises because the ak uses a keplerian orbital parameterisation , compared to the geodesic parameterisation used in the nk . for an equatorial orbit with semi - latus rectum @xmath27 and eccentricity @xmath28 around a @xmath29 black hole of spin @xmath30 ,
the azimuthal frequency of the nk waveform is @xmath31mhz compared to @xmath32mhz for the ak .
these orbits will therefore be one cycle out of phase within @xmath33 hours .
ak waveforms will thus not be particularly _ faithful _ templates .
the problems can be mitigated by adjusting the orbital parameters of the ak waveform to improve the match with the nk , and the ak waveforms do capture the main features of true emri waveforms .
for this reason , they may be quite _ effectual _ templates , but this has not yet been properly assessed . in the future ,
the effectualness of the ak waveforms as detection templates will be investigated by using banks of ak templates to search for more accurate nk or tb waveforms embedded in noise .
the idea behind the numerical kludge ( nk ) waveforms is to combine an exact particle trajectory ( up to inaccuracies in the phase space trajectory and conservative radiation reaction terms ) with an approximate expression for the gw emission . by including the particle dynamics accurately ,
we hope to capture the main features of the waveform accurately , even if we are using an approximation for the waveform construction .
the computation of nk waveforms proceeds in three steps .
the first is to construct an inspiral trajectory in `` phase space '' that is , the space of constants @xmath34 ( energy ) , @xmath35 ( axial angular momentum ) and @xmath36 ( carter constant ) which characterizes kerr black hole orbits ( up to initial conditions ) .
the construction of the phase space trajectory has already been described in previous work @xcite . in this paper
we shall ( in most cases ) ignore the evolution of orbits due to radiation reactions .
the procedure for waveform construction including orbital evolution is identical to that given here .
the second step is to integrate the kerr geodesic equations along the inspiral trajectory and hence obtain the boyer - lindquist coordinates of the inspiralling object as a function of time @xcite .
the final step is to construct a gravitational waveform from this inspiral trajectory .
we do this by identifying the boyer - lindquist coordinates ( @xmath37 ) with spherical polar coordinates in flat - space and then evaluating a flat - space emission formula for the corresponding flat - space source orbit .
the idea of coupling a weak field formula with fully relativistic motion first appeared in papers by sasaki and ruffini @xcite and was then investigated more thoroughly for a test - body orbiting a schwarzschild black hole by tanaka et al.@xcite .
more recently , gair et al .
@xcite looked at semi - relativistic fluxes for schwarzschild orbits and derived analytic expressions for the fluxes from arbitrary orbits .
their focus was on highly eccentric orbits relevant to the capture problem .
in all these cases , the authors were interested in computing semi - relativistic fluxes , rather than waveforms .
while our focus is on waveforms rather than fluxes , in section [ results_flux ] we do calculate semi - relativistic fluxes for kerr orbits , as these results have not yet appeared in the literature . in the remainder of this section
, we describe the two stages of waveform generation ( ignoring backreaction ) :
( i ) the computation of a trajectory , and ( ii ) the computation of a gravitational waveform from an arbitrary trajectory .
the first step in constructing a numerical kludge waveform is to compute the trajectory that the inspiralling body follows in the boyer - lindquist coordinates of the kerr spacetime of the central black hole .
ignoring radiation reaction , this path is a kerr geodesic .
geodesic motion in the kerr space - time is well - known @xcite and is governed by a set of first - order differential equations : & = & , + & = & , + & = & v _ , + & = & v_t , where the various `` potentials '' are defined by v_r & = & ^2 - , + v _ & = & q -^2 , + v _ & = & - ae + , + v_t & = & a ( l_z - ae^2 ) + . here , @xmath38 , and @xmath39 .
the constants @xmath34 , @xmath35 , @xmath36 are the three first integrals of the motion : @xmath34 is the orbital energy ; @xmath35 is the projection of the orbital angular momentum along the black hole s spin axis ; and @xmath36 is known as the `` carter constant '' .
this last constant is the relativistic generalization of the `` third integral '' used to separate the equations which describe orbits in an axisymmetric gravitational potential ( a result which is particularly well - known in the literature describing orbits in galactic potentials @xcite ) . in the spherical ( i.e. , schwarzschild ) limit
, @xmath36 reduces to the square of the orbital angular momentum projected into the equatorial plane ; see @xcite and references therein for discussion .
for a given @xmath34 , @xmath35 and @xmath36 , the roots of @xmath40 determine the turning points of the radial motion the periastron @xmath41 , and apastron @xmath42 .
from these , one can define an orbital eccentricity @xmath43 , and semi - latus rectum @xmath44 , using the conventional keplerian definitions [ pedef ] r_p= , & & r_a= , + p = , & & e=. we also replace the carter constant by an `` inclination angle '' , defined by ^2= .
[ iotadef ] to aid numerical integration , one can work in terms of two angular variables , @xmath45 and @xmath46 , instead of @xmath47 and @xmath48 .
we define @xmath45 by the equation @xmath49 we define @xmath46 by the equation @xmath50 , where @xmath51 is given by @xmath52 + q,\ ] ] with @xmath53 . expanding the radial potential as @xmath54 we find evolution equations for @xmath45 and @xmath46 of the form @xmath55^{\frac{1}{2}}\,\left[(p - r_{4}(1 + e ) ) + e(p - r_{4}(1 + e)\cos\psi)\right]^{\frac{1}{2}}}{\left[\gamma + a^{2}\,e\,z(\chi)\right](1 - e^{2 } ) } , \label{dpsidt } \\
\frac{\rmd \chi}{\rmd t } & = & \frac{\sqrt{\beta\,\left[z_{+ } - z(\chi)\right]}}{\gamma + a^{2}\,e\,z(\chi ) } \label{dchidt } \\ \nonumber { \rm where } \hspace{0.1in}\gamma & = & e\left[\frac{\left(r^2 + a^2\right)^{2}}{\delta } - a^2 \right ] -\frac{2\,m\,r\,a\,l_{z}}{\delta } .\end{aligned}\ ] ] in terms of the variables @xmath56 , @xmath45 and @xmath46 , the geodesic equations are well behaved at the turning points of the motion , which facilitates numerical integration ( as discussed in @xcite ) .
although we mostly ignore it here , it is easy to include radiation reaction in this prescription .
to do so , one first computes an inspiral trajectory through phase space by writing [ inspeqns ] & = & f_e ( a , m , m , p , e , ) , + & = & f_l ( a , m , m , p , e , ) , + & = & f_q ( a , m , m , p , e , ) .
the functions @xmath57 , @xmath58 and @xmath59 are derived from well - known post - newtonian results , augmented by additional corrections .
the leading order part of @xmath57 is given in a later section of this paper , eq .
( [ ghkedot ] ) .
( for the full expressions see ref .
since the parameters @xmath60 , @xmath26 and @xmath61 are constant , and @xmath44 , @xmath43 and @xmath62 are directly related to @xmath34 , @xmath35 and @xmath36 via eqs .
( [ pedef])-([iotadef ] ) , equations ( [ inspeqns ] ) can be integrated to give the phase space trajectory .
that is , the values of @xmath34 , @xmath35 and @xmath36 ( or equivalently @xmath44 , @xmath43 and @xmath62 ) are given as functions of time . to obtain the trajectory of the inspiralling particle , these time dependent expressions
can be substituted into the right hand sides of eqs .
( [ geoeqphi ] ) , ( [ geoeqt ] ) , ( [ dpsidt ] ) and ( [ dchidt ] ) . the resulting set of odes can then be integrated to give the inspiral trajectory . in the following ( with one exception )
, we will only consider waveforms from geodesic trajectories ( setting @xmath63 ) , and refer the reader to refs .
@xcite for more details on the construction of inspirals .
once a trajectory has been obtained in this manner , one constructs an `` equivalent '' flat - space trajectory by projecting the boyer - lindquist coordinates onto a fictitious spherical polar coordinate grid , defining the corresponding cartesian coordinate system and pretending that these new coordinates are true flat - space cartesian coordinates : x & = & r , + y & = & r , + z & = & r .
we use the resulting flat space trajectory as input to a wave generation formula .
this is a `` bead on a wire '' prescription by putting the trajectory in flat space , we remove the background that is causing the curvature of the geodesic path and hence we are forcing the particle to move along a curved path like a bead moving on a wire .
this leads to obvious inconsistencies in the approach e.g. , the non - conservation of the flat - space energy - momentum tensor of the particle motion since we are not including the energy - momentum of `` the wire '' along which the particle moves .
having constructed the particle orbit in our pseudo - flat - space , we now apply a flat - space wave generation formula .
consider the weak - field situation and write down the spacetime metric as @xmath64 , where @xmath65 is the flat metric and @xmath66 are small perturbations .
the _ trace - reversed _
metric perturbation is defined as @xmath67 , where @xmath68 . imposing the lorentz gauge condition @xmath69 , the linearized einstein field equations can be written as @xmath70 in which @xmath71 denotes the usual flat space wave operator and the effective energy - momentum tensor @xmath72 satisfies @xmath73 here a comma subscript denotes partial derivative ( @xmath74 ) .
note that our source conservation equation uses a partial rather than a covariant derivative .
this is because we would hope , in a self - consistent approach , to choose coordinates so that the energy momentum tensor is flat - space conserved . finally ,
when observing gws at large distances , we are really only interested in the transverse and traceless parts of the spatial components of @xmath75 ; a projection of these components will be necessary .
taking coordinates centered at the black hole , we denote the position of the observer by @xmath76 and the position of the particle by @xmath77 ) .
the wave equation ( [ waveq ] ) has the familiar retarded time solution @xmath78 the additional coordinate @xmath79 is the integration variable , which goes over every possible space location where the effective energy - momentum tensor @xmath80 is nonzero .
if the source motion is only negligibly influenced by gravity , then @xmath72 may be taken to equal @xmath81 , the energy - momentum tensor of the matter source . in ref .
@xcite , press derived a formula valid for extended , fast motion sources .
this was obtained by substituting eq .
( [ emcons ] ) into eq .
( [ retint ] ) repeatedly .
the result is @xmath82_{t'=t-|\bx-\bxd| } \rmd^{3 } x ' , \label{press}\ ] ] in which @xmath83 and @xmath84 . in the slow motion limit
, the press formula reduces to the usual quadrupole formula ( hereafter , an overdot denotes a time - derivative ) @xmath85_{t'=t - r},\ \label{quad}\end{aligned}\ ] ] where @xmath86 is the source s mass quadrupole moment . including the next order terms ( the mass octupole and current quadrupole moments of the source , denoted @xmath87 and @xmath88 respectively ) , we obtain the quadrupole - octupole formula ( see refs .
@xcite for details ) , @xmath89_{t'=t - r},\ \label{octup}\end{aligned}\ ] ] with @xmath90 note that in both eqs .
( [ quad ] ) and ( [ octup ] ) , the retarded time is @xmath91 instead of the more complicated expression appearing in ( [ press ] ) .
if desired , it is a straightforward ( but increasingly tedious ) task to include more terms in the slow - motion expansion of ( [ press ] ) . in the present work we shall make use of eqs .
( [ quad ] ) , ( [ octup ] ) and the full press formula ( [ press ] ) .
the waveform in the standard transverse - traceless ( tt ) gauge is given by the tt projection of the above expressions .
we define an orthonormal spherical coordinate system via _
r = , _ = , _ = .
the angles @xmath92 denote the observation point s latitude and azimuth respectively .
the waveform in transverse - traceless gauge is then given by @xmath93 with h^ & = & ^2 + h^zz^2 - , + h^ & = & + , + h^ & = & h^xx^2 - h^xy + h^yy^2 .
the usual `` plus '' and `` cross '' waveform polarizations are given by @xmath94 and @xmath95 respectively
. the expressions ( [ press ] ) , ( [ quad ] ) and ( [ octup ] ) are valid for a general extended source in flat - space . if we specialize to the case of a point - mass @xmath16 moving along a trajectory @xmath96 , we know that the energy - momentum tensor in flat spacetime is given by @xmath97 where @xmath98 is the proper time along the trajectory .
it is related to the particle s coordinate time by @xmath99 , where @xmath100 . on the right hand side of eq .
( [ emt ] ) is a term @xmath101 .
expressions ( [ quad ] ) and ( [ octup ] ) are slow - motion expansions and at the order of these expansions we should replace @xmath102 by @xmath103 for consistency .
the press formula ( [ press ] ) is a fast - motion expression , and therefore we do include this term .
the presence of a @xmath23-function in @xmath81 facilitates the simplification of the various moments in eqs .
( [ quad ] ) and ( [ octup ] ) : @xmath104 here @xmath105 .
the press formula similarly simplifies to @xmath106_{t'_p = t- |\bx -\bx'_p| } .
\label{pressexp}\ ] ] the square bracketed expression is to be evaluated at a time @xmath107 given implicitly by @xmath108 . in a numerical implementation
, we evaluate the expression as a function of the time @xmath107 along the particle s path . in order to obtain a time series that is evenly spaced in @xmath109 and
compute the rhs of eq .
( [ pressexp ] ) by finite - differencing , we adjust the spacing of the sampling at the particle , @xmath110 , such that @xmath111 with a delta function source ( [ emt ] ) , the retarded - time solution ( [ retint ] ) can be evaluated directly @xcite : @xmath112_{t'_p = t-|\bx-\bx'_p|}. \label{retexp}\ ] ] naively , one could suppose that ( [ retexp ] ) will perform better than ( [ press ] ) , since it is derived using only one of the two ( invalid ) flat space equations , rather than both .
in fact , we find that the retarded integral expression ( [ retexp ] ) performs much worse than either ( [ quad ] ) , ( [ octup ] ) or ( [ press ] ) when compared to tb waveforms .
the reason appears to be that the manipulations which lead to the quadrupole , quadrupole - octupole , and press formulae ensure that the actual source terms mass motions are on the right hand side .
this is to be contrasted with the retarded integral expression ( [ retint ] ) , which identifies the dominant gw on the left hand side with weak spatial stresses on the right .
we must emphasize that the nk prescription is clearly inconsistent we are binding the particle motion to a kerr geodesic while assuming flat spacetime for gw generation and propagation .
this is manifested by the fact that the energy - momentum tensor ( [ emt ] ) is not flat space conserved , @xmath113 .
however , the spirit of this calculation is _ not _ a formal and consistent approximation to emri waveforms ; it is rather a `` phenomenological '' approach which takes into account those pieces of physics we believe are the most crucial in particular , the exact kerr geodesic motion . by including the exact source trajectory ,
we ensure that the spectral components of the kludge waveforms are at the correct frequencies , although their relative amplitudes will be inaccurate .
as we shall see , this line of thinking is validated _ post facto _ by the remarkable agreement between the kludge and tb waveforms ( see section [ results ] ) .
it is important to underline the physical assumptions that have been made in the derivation of ( [ press ] ) , ( [ quad ] ) and ( [ octup ] ) , in order to understand their generic limitations .
first of all , the assumed absence of any background gravitational field means that our kludge waveforms are unable to capture any features related to back - scattering .
this effect is known to first appear at 1.5pn [ i.e. @xmath114 level ( see , e.g. , @xcite ) .
such `` tails '' of waves are particularly prominent in the strong - field tb equatorial `` zoom - whirl '' waveforms @xcite and in the waveforms from plunging or parabolic orbits @xcite . in all these cases
, the hole s quasinormal mode ringing adorns the emitted signal .
the slow - motion nature of expressions ( [ quad ] ) and ( [ octup ] ) suggests that they might be bad models for waveforms generated by orbits venturing deep inside the central bh s spacetime .
the rich multipole structure of the true waveform from such orbits is poorly reproduced , as any slow - motion approximation essentially truncates the sum over multipoles .
the press formula ( [ press ] ) includes contributions at all multipoles and so might be expected to handle these contributions quite well . however , it turns out that it does not perform that much better than the quadrupole - octupole waveform in this regime .
while it includes contributions at all multipoles , the lack of background curvature in the waveform model means that the amplitude of the higher modes is much lower than expected for true emri waveforms .
fortunately , these two deficiencies become important for the same class of orbits those that allow the body to approach very close to the black hole . as a rule of the thumb we shall find that nk waveforms are reliable ( in terms of the overlap discussed in the next section ) as long as the closest orbital approach ( periastron ) is @xmath115 .
we now give a brief description of the measure used for the quantitative comparison between nk and tb waveforms .
the main motivation for the computation of accurate emri waveforms is to carry out matched filtering for detection of the gws . for the purpose of signal detection and parameter estimation ,
a bank of templates is constructed which covers the desired parameter space with sufficient resolution . the detector output
is then filtered through each template .
the measured strain amplitude in the detector , @xmath116 , consists of a ( possibly present ) signal @xmath117 , and the detector noise @xmath118 .
we define the fourier representation of these time series as @xmath119 the signal - to - noise ratio ( snr ) can be expressed in terms of an inner product defined on the vector space of possible signals . given two vectors ( time series ) , @xmath120
, we define the overlap @xmath121 by the equation @xcite @xmath122 where @xmath123 is the one - sided noise power spectral density ( psd ) ( in this case , the noise psd for the lisa detector ) and an asterisk denotes complex conjugation . considering @xmath124 as a template with parameters @xmath125 we can approximate the snr by @xcite @xmath126 = \frac{\langle(s|h)\rangle_n}{\sqrt{\langle(n|h)^2\rangle_n } } = \frac{(s|h)}{\sqrt{(h|h)}}\;.\ ] ] the notation @xmath127 means to ensemble average the function @xmath128 over all possible noise realizations @xmath129 .
the psd in the denominator of eq .
( [ inprod ] ) serves to suppress those frequency components of the signal at which the detector noise is large .
the main tool which will be used in this paper is not the snr , but rather the overlap function @xmath130 .
the overlap is defined as an inner product between two normalized vectors / signals : @xmath131 the normalization is chosen so that @xmath132 .
the overlap can be regarded as the inner product between two unit vectors ; it varies within @xmath133 $ ] .
the overlap is equal to @xmath103 if the two waveforms are identical , and it equals zero if the two waveforms are orthogonal ( for example , cosine and sine signals ) .
the overlap is an appropriate measure of `` goodness - of - fit '' since we are interested in knowing how well the nk approximate the behavior of tb emri waveforms . in this context
, it is important to include the noise properties of the detector since it is no problem if a template has poor performance at frequencies where detector noise is large . for a fair comparison , we should choose the signal ( tb waveform ) and template to have the same parameters .
in practice , the start times of our signals and templates were slightly mismatched , so we allow for maximisation over starting time ( `` time of arrival '' ) .
this maximisation is commonly done in gw data analysis , and can be accomplished very efficiently in the fourier transform , since the time offset corresponds to a phase shift in the frequency domain .
accordingly , we modify the inner product @xmath121 slightly : @xmath134 where @xmath109 corresponds to the time difference .
we find that the maximum usually occurs with an time offset close to zero ( typically a few bins ) .
we conclude our discussion of the overlap by describing our approximation to the lisa noise @xmath123 .
we have used an analytic approximation to the numerically generated sensitivity curve given in ref .
the agreement between these two curves is excellent as clearly illustrated in fig .
[ s_h(f ) ] .
our approximate @xmath123 function can be easily calculated according to the following simple prescription .
define @xmath135 , where @xmath128 is frequency and @xmath136 sec is the light travel time down one of lisa s arms .
for @xmath137 we set @xmath138 while for @xmath139 , @xmath140 where @xmath141\;. \label{r}\end{aligned}\ ] ] note that lisa s characteristics are incorporated in the light travel time @xmath98 and in the numerical constants in the above expressions , so a similar mission but with different noise characteristics would still be described by the functional form defined by eqns .
( [ s1]-[r ] ) .
we have also included noise from a confused population of galactic white dwarf binaries following the prescription outlined in @xcite . for lisa ;
black curve : numerical curve as generated in @xcite , red curve : analytic approximation used in this paper , see text for details ) . ]
in this section we compare nk waveforms ( using the overlap function defined in the previous section ) to a variety of tb waveforms from inclined - circular , equatorial - eccentric and some generic kerr orbits .
the tb waveforms are treated here as `` exact '' ( though this is not strictly the case as discussed in section [ inventory ] ) .
our comparison rule is that the waveforms are generated using the same kerr orbit with identical initial conditions .
we label these orbits by the triad of orbital elements @xmath142 defined in equations ( [ pedef])([iotadef ] ) .
figures [ equat4 ] , [ circ3 ] , and [ gener ] show time - domain waveforms for a selection of orbits in the moderate and strong field .
the nk waveforms plotted in these figures were generated by using the quadrupole - octupole formula ( [ octup ] ) .
the results for `` plus '' and `` cross '' polarizations are essentially the same , so in the figures we show only the `` plus '' polarization . for all of the figures , we have assumed an observation point located at @xmath143 and @xmath144 or @xmath145 ( as indicated )
. direct visual inspection of the waveforms gives some indication of how well the nk and tb waveforms match . in each of the cases
illustrated , the kludge waveforms manage to capture the overall wave pattern , and for the orbits with @xmath146 they almost exactly match the tb waveforms . for ultra - relativistic orbits ( e.g. , with @xmath147 and @xmath148 ) ,
the finer structure in the tb waveforms is clearly not reproduced by the quadrupole - octupole kludge waveforms . as we have discussed , these features are the imprints of higher multipole components in the radiation which are amplified by back - scattering , and thus are not expected to appear in the nk waveforms .
waveforms generated using the press formula ( not shown ) do have some finer features due to the presence of higher multipole components , as we might have hoped , but they are not nearly as complicated as the structure of the tb waveforms .
one also notices that for certain parts of the waveforms ( e.g. , fig . [ circ3 ] )
, there is a disagreement in the amplitude while the phase is accurately reproduced .
this amplitude discrepancy is periodic , i.e. , the points where the amplitude is poorly reproduced occur at regular intervals .
this suggests that the kludge waveforms are missing some periodic components , as we might expect since they are truncated expansions in multipole moments .
indeed , the amplitude discrepancy is less pronounced in the press waveform , which includes contributions at all multipoles . for the purposes of data analysis ,
it is important to have templates with a high overlap with the true signals , and to achieve that it is much more important that the waveform phase is reproduced rather than the waveform amplitude . the waveform phase is determined by the orbit generating the gravitational radiation .
the fact that the kludge waveforms are based on true geodesic orbits is presumably the reason that we find , _ post facto _ , such impressively high overlaps with tb waveforms and especially for those from circular - inclined orbits .
a comprehensive set of data for the overlaps between nk and tb waveforms is given in tables [ o - lap_equat ] , [ o - lap_circ ] and [ o - lap_generic ] in the appendix .
these were computed using the overlap function described in section [ overlaps ] and assuming a central black hole mass of @xmath149 .
this was chosen since preliminary event rate estimates suggest the inspirals of @xmath150 bhs into @xmath151 smbhs will dominate the lisa detection rate @xcite .
these tables indicate that if the orbital periastron @xmath152 , the overlap between tb waveforms and both the quadrupole - octupole and press waveforms stays above @xmath153 .
we also find that both these expressions have better performance than the pure quadrupole waveforms ( [ quad ] ) , but there is little difference between quadrupole - octupole and press waveforms .
the kludge and tb waveforms begin to deviate significantly for strong - field , ultra - relativistic orbits with @xmath154 , with the overlap dropping to @xmath155 for orbits that come very close to the horizon .
disappointingly , the press waveforms do not seem to do much better than the quadrupole - octupole waveforms in this strong field regime , despite the inclusion of additional multipole components .
the press waveforms include only the `` direct '' contribution to the higher multipoles , i.e. , the piece that arises from fast motion in flat spacetime .
in true emri gws , the higher multipole contributions are significantly enhanced by `` tail '' effects , i.e. , the backscattering of radiation from the background geometry .
since we are using the press formula in flat space , we do not include this back - scattering enhancement .
this is presumably why using the kludge press waveforms does not significantly improve the overlap with tb waveforms .
the press waveforms do perform consistently better for circular orbits and weak - field eccentric orbits , but the difference between the two approaches is usually small ( with the exception of a few cases which we discuss in more detail later ) .
we conclude that the quadrupole - octupole waveform model is sufficient and there is not much gain from using the computationally more intensive press formula . to summarize
, we find that kludge waveforms are accurate and very quick to generate substitutes for tb waveforms for all orbits with periapse @xmath15 . in schwarzschild ,
the separatrix between bound and plunging orbits is at @xmath156 , so the kludge waveforms will be accurate throughout any inspiral with eccentricity at plunge @xmath157
. computations of inspirals into schwarzschild black holes @xcite indicate that , in many cases , the residual eccentricity at plunge will be small , so that kludge waveforms will be suitable for the majority of schwarzschild inspirals . for retrograde orbits around kerr black holes ( @xmath158 ) ,
the minimal periapse is at even larger radii , so that even weaker restrictions must be imposed on the eccentricity at plunge for the kludge to be valid .
in contrast , for prograde orbits ( @xmath159 ) , an increased black hole spin allows stable orbits to exist much deeper in the strong field . for @xmath160 ,
the separatrix for circular orbits is at @xmath161 for orbits with inclinations of @xmath162 respectively ; for orbits with a plunge eccentricity of @xmath163 , it is at @xmath164 for the same inclinations . as @xmath165 , the separatrix of equatorial orbits decreases even more , asymptotically approaching @xmath166 @xcite .
kludge waveforms are not particularly good in this regime , with overlaps @xmath167 ; fortunately , this corresponds to a comparatively small region of parameter space .
the evolution proceeds through this region very quickly , so we do not lose much signal - to - noise ratio by failing to match the waveforms in this region .
if the overlap between a given signal and the best - fit template in a search bank is less than @xmath103 , this leads to a decrease in the maximum distance to which that signal can be detected , and a corresponding reduction in event rate . for the purposes of detection ,
overlaps as low as @xmath168 might well be considered good enough , if the astrophysical event rate is sufficiently large @xcite . however , for the purposes of parameter estimation , much higher overlaps ( @xmath169 ) will be required in general .
it is clear from the results in this paper that this is only partially achievable by the existing family of kludge waveforms .
nonetheless , these waveforms might be useful for lisa data analysis as search or detection templates over some ( perhaps a large part ) of the astrophysically relevant portion of the @xmath170 parameter space .
the waveforms may also provide sufficiently accurate estimation of the source parameters ( in certain regions of parameter space ) that they could be used as the first stage in a hierarchical search .
the purpose of such a search would be to identify `` interesting '' regions of parameter space for follow up with more accurate waveforms .
this is the main conclusion of this paper .
however , the regions where kludge waveforms are good enough must be identified more carefully by comparison to accurate _
inspiral _ waveforms .
as we have discussed earlier , the flat space emission formula used in the construction of the nk waveforms ignores all effects of scattering from the background curvature .
these `` tail '' terms make a significant contribution to the waveform structure , and build up over the course of an inspiral .
although we have found good overlaps with geodesic waveforms and circular - inclined inspiral waveforms ( see section [ inspcomp ] ) , comparisons to inspirals of eccentric - inclined orbits are required to properly assess the importance of including the tail terms .
accurate , self - force waveforms for such orbits will not be available for a few years and only then will it be possible to firmly demarcate the regime of usefulness of the present , or further improved , nk waveforms . or @xmath145 .
orbital parameters are listed above each graph .
the waveforms are scaled in units of @xmath171 where @xmath172 is the radial distance of the observation point from the source and @xmath16 is the test - body s mass .
the x - axis measures retarded time ( in units of @xmath26 ) and we are showing the ` + ' polarization of the gw in each case .
the overlaps between the nk and tb waveforms are @xmath173 going from the top figure down . ] .
orbital parameters are listed above each graph .
the waveforms are scaled in units of @xmath171 where @xmath172 is the radial distance of the observation point from the source and @xmath16 is the test - body s mass .
the x - axis measures retarded time ( in units of @xmath26 ) and we again show the plus polarization of the gw .
the overlaps between the nk and tb waveforms are @xmath174 for the top figure and @xmath175 for the bottom figure . ] where @xmath172 is the radial distance of the observation point from the source and @xmath16 is the test - body s mass .
the x - axis measures retarded time ( in units of @xmath26 ) .
the overlaps between the nk and tb waveforms are @xmath176 and @xmath177 for the top and bottom figures respectively . ] to better understand the overlaps quoted in the tables we consider here the integrand of eq .
( [ max_t ] ) for the value of @xmath109 which maximizes the overlap : @xmath178 in figure [ nks ] we plot this function for the generic orbit @xmath179 and @xmath160 . for the signal
@xmath180 we use the tb waveform and correlate it with templates @xmath181 . as templates we use the tb waveform itself ( black solid line ) , the press waveform ( squares ) , the quadrupole - octupole nk waveform ( crosses ) and the quadrupole nk waveform ( circles ) .
we have deliberately chosen a case for which including higher harmonics significantly ( @xmath182 ) improves the overlap , especially in the presence of white dwarf confusion noise ( see table [ o - lap_generic ] ) .
one can see that the main contributions to the overlap come from several dominant harmonics .
for a circular equatorial orbit the main harmonic would correspond to twice the orbital frequency , but in general the main harmonics depend on eccentricity @xcite . besides the harmonics coming from the azimuthal motion there are many additional components coming from beating between harmonics of the fundamental frequencies of the @xmath56- , @xmath48- and @xmath47-motion : @xmath183 ( see @xcite for a fourier decomposition of the orbital motion ) . depending on the mass @xmath26 of the central bh , the dominant harmonics may lie in the most sensitive part of lisa s frequency range or they can be suppressed by confusion noise . in the latter case
the higher harmonics , although smaller in amplitude , are effectively enhanced by the inverse power spectral density and therefore play an important role in snr accumulation . looking at the inlaid box in figure [ nks ] we can compare the quadrupole - octupole and press waveforms at high frequency . here
, the press waveform does perform better
e.g. , around 3mhz the quadrupole - octupole nk waveform has failed to reproduce some harmonics in the tb signal , whereas the press waveform does so pretty well .
we will discuss the frequency representation of the signals further in the next subsection . as described earlier , most of the existing tb waveforms describe gw emission from geodesic orbits , which is why we have focused on these for the comparisons above .
however , tb waveforms have also been constructed for the inspiral of circular - inclined orbits @xcite . to illustrate the applicability of the kludge to the detection of realistic ,
inspiralling emri waveforms , we here compare a kludge inspiral waveform to a tb circular - inclined inspiral waveform generated using the method described in @xcite .
the kludge inspiral trajectory is constructed using the flux formulae described in @xcite , and then the particle trajectory is computed as discussed in section [ insptraj ] .
we use the quadrupole - octupole formula ( [ octup ] ) to generate a kludge waveform from this trajectory .
a visual comparison is shown in figure [ inspcom ] . there
we plot both the inspiral trajectory @xmath184 and the gravitational waveform @xmath185 .
one can see an excellent agreement up to the last several cycles , where we would not trust either type of waveform anyway . we have used a large mass ratio ( @xmath186 ) to compute the waveforms in order to speed up the orbital evolution .
we should point out that the comparison is not improved if we use press waveforms instead of quadrupole - octupole waveforms .
we may quantify the agreement in the waveforms by again using the overlap function , which we plot as a function of truncation time in figure [ inspolap ] for three different masses of the central black hole .
one can see that for the majority of the inspiral we obtain quite high overlap , but it decreases when we include the final stages of the inspiral . for the inspiral in figure [ inspcom ] , the overlap ( when @xmath187 and in presence of the white dwarf background ) up until radii of @xmath188 is @xmath189 respectively .
the overlap over the whole inspiral is still greater than @xmath190 .
an overlap of @xmath190 corresponds to a factor of @xmath191 reduction in event rate .
however , overlaps of @xmath192 or @xmath193 correspond to a reduction in event rate of only @xmath194 or @xmath195 respectively . a significant amount of signal to noise ratio accumulates over the early stage of inspiral , and
so it may well be that kludge inspiral waveforms can be used to efficiently detect and track events until they reach @xmath196 .
alternative techniques could then be employed for detection during the latter stages of inspiral .
the mass of the central bh determines the frequency range through which the inspiral occurs .
one could also evaluate the spectrum of the signal as a function of the dimensionless frequency @xmath197 the effect of changing @xmath26 is then to shift the psd @xmath123 right as @xmath26 is increased or shift it left as @xmath26 is decreased . in figure [ inspint ]
we again plot the overlap integrand ( [ integrand ] ) for @xmath187 .
we also overplot the cumulative overlap @xmath198 and a scaled psd . by shifting @xmath123 on the plot left
/ right one can see how the central mass affects the amount each harmonic contributes to the snr / overlap .
the higher harmonics are modeled less accurately by nk waveforms and this leads to an overall drop in the overlap for @xmath199 .
however , the majority of the high frequency contribution to the waveforms comes from the end of the inspiral which we do not reproduce very accurately , rather than from higher harmonics emitted during the earlier , moderately relativistic part of the orbit . we should also mention that there is a problem with spectral leakage @xcite ( due to the large dynamic range ) in the estimation of gw spectra
this is not a big problem on its own but for high @xmath26 this leakage ( at high frequencies ) is amplified by the inverse power spectral density and leads to an erroneous result . in order to reduce this effect
we have applied a window function @xmath200 + \tanh\left [ a(t - t - \kappa t ) \right ] \right)\ ] ] to the template and signal for @xmath199 and in the presence of confusion noise .
the parameters @xmath201 and @xmath202 govern the steepness and cut - off point of the window ; @xmath203 is the waveform s duration .
this function reduces the effect of the end of inspiral on the overlap and leads to an artificial increase in the overlap by a few percent . as mentioned above , a significant amount of signal to noise ratio accumulates over the early stages of the inspiral . to illustrate this we show in figure [ inspsnr ] the accumulation of snr as a function of time . to generate this plot ,
we have assumed the tb waveform is the signal and have used the nk waveform , truncated at different times , as the search template .
one can see that we can recover up to 85% of the maximum snr ( i.e. , the snr if the template and signal were identical ) .
including the last few waveform cycles we see a drop in the snr ( @xmath204% ) due to the mismatch in the waveforms at the end of the inspiral . employing press waveforms
helps to improve overlaps further , but for ultra - relativistic orbits the omitted tail contribution to the waveform becomes increasingly significant .
it is possible to construct expressions for the flux of energy and angular momentum carried by the kludge gws , and use these as estimates of the energy and angular momentum lost from the orbit in a true inspiral .
the energy and angular momentum content of a tt gw field , @xmath205 , propagating in flat space at large distances from the source is determined from the isaacson energy momentum tensor of the wave field @xcite @xmath206 integration of this expression gives the energy and angular momentum loss rates due to gw emission ( equations ( @xmath207 ) and ( @xmath208 ) of @xcite ) as @xmath209 using the quadrupole - octupole formula to generate @xmath205 , these expressions may be written in terms of the multipole moments of the source as @xmath210 in these expressions @xmath211 and bracketed numbers in superscripts denote the number of time derivatives , e.g. , @xmath212 . if the kludge waveform is computed using the quadrupole formula ( [ quad ] ) , only the leading order @xmath213 terms remain .
if a press waveform is used , there is an infinite sum of multipole components , and therefore expressions ( [ genedot][genjdot ] ) must be employed directly .
the angle brackets in expressions ( [ genedot][improvjdot ] ) mean `` average over several gravitational wavelengths '' .
this is achieved by taking a time average of the instantaneous flux expression inside the angle brackets , i.e. , @xmath214 where @xmath203 is an appropriate averaging time .
it is important to ensure that this averaging time is long enough . for circular
equatorial orbits , the symmetries of the system ensure that the instantaneous and averaged fluxes are identical . for an eccentric equatorial orbit ,
there is only one periodicity in the flux the period of the radial motion . thus integrating over one radial period is sufficient to give an accurate average flux . for generic orbits ,
the averaging is more complicated since the periods of the motion in the radial and @xmath48 directions are in general different and non - commensurate .
the averaging time therefore has to be long enough to encompass several periods of both motions .
we illustrate this in fig .
[ fig1 ] by plotting @xmath215 as function of @xmath203 for a generic orbit with parameters @xmath216 and @xmath217 .
the flux has been computed using the pure quadrupole waveforms ( [ quad ] ) .
it is clear from fig .
[ fig1 ] that the averaged flux converges over time , and this convergence is more rapid for the lower eccentricity orbit .
in all subsequent flux calculations , we took @xmath218 ( which corresponds to @xmath33 days for @xmath219 ) . in figure [ fig2 ]
we show the angular distribution of the gravitational radiation from an emri orbit of given @xmath44 and @xmath43 , but for several different orbital inclinations , @xmath62 .
this picture is more or less typical of the majority of orbits .
the variation of the energy flux with the sky position of the observer is indicative of beaming .
this is strongest for equatorial orbits and reduces as the orbital inclination is increased .
equatorial orbits are restricted to a single plane , while inclined orbits wander through more of the spacetime .
this wandering averages out the beaming over the sky .
the larger the orbital inclination , the more of the phase space the body explores , the more averaging occurs and the more homogeneous the sky distribution of the energy flux . in tables [ tab3 ] & [ tab4 ] we show flux data for a series of orbits and black hole spins .
we tabulate the flux computed via solution of the teukolsky equation , and fluxes computed using quadrupole kludge waveforms and equation ( [ improvedot ] ) .
we quote only the quadrupole results since the difference from using either the quadrupole - octupole or press waveforms is quite small . as we mentioned in section [ inventory ] , tanaka et.al .
@xcite and more recently gair et al .
@xcite computed approximate fluxes for schwarzschild orbits based on the quadrupole waveform approximation ( [ quad ] ) . for a random choice of orbits
we have found excellent agreement with their results . for further comparison
, we also tabulate the `` kludge '' fluxes that have been used to construct inspirals and inspiral waveforms for scoping out lisa data analysis .
the difference between these and our current results is a measure of the inconsistency in the kludged inspiral waveforms . in constructing an inspiral waveform ,
one set of expressions is used to evolve the orbit ( given in @xcite ) , and a different prescription ( the one described here ) is used to generate the waveform .
this means that the energy carried by the kludge gws is not equal to the energy lost by the orbit that is supposedly emitting these gws .
this leads to inaccuracies when the kludge waveforms are used to compute snrs @xcite . in the original prescription for kludge
inspirals @xcite , the phase space trajectories were generated using the leading order post - newtonian results of ryan @xcite , evaluated for relativistic orbital parameters .
ryan s expression for the energy flux is @xmath220 \label{ghkedot}\ ] ] where the @xmath43-dependent coefficients are , @xmath221 in a more recent version of the kludge @xcite , this prescription for the evolution of the orbital parameters was improved .
the new expression for @xmath215 is complicated and we do not quote it here , but we also tabulate those results in tables [ tab3 ] & [ tab4 ] , under the heading `` gg '' .
we note that in true black hole spacetimes , energy and angular momentum are lost into the horizon as well as to infinity .
the teukolsky equation provides a prediction for both the flux of radiation at infinity and the flux down the black hole horizon . when using eqs .
( [ genedot])([improvjdot ] ) , we are only computing the energy and angular momentum carried to infinity in the kludge gws .
this should therefore only be compared to the teukolsky flux at infinity .
the discrepancy between these numbers is then a measure of the error associated with using the kludge waveforms to estimate signal to noise ratios for true inspirals .
the tb fluxes quoted in the tables are accordingly just the infinity piece of the fluxes .
when computing inspirals , which was the purpose for which the `` gg '' expressions were derived , the energy and angular momentum lost by the orbit should include the horizon piece of the flux .
this should be born in mind when considering the flux tables . however , the relative contribution of the horizon flux is only significant in the extremely strong field regime where both the `` gg '' expressions and the kludge fluxes cease to be valid . in general , we find that the fluxes computed from the kludge waveforms agree less well with teukolsky results than do the `` gg '' fluxes
. the kludge fluxes do improve over `` gg '' results for some retrograde orbits , and for some orbits with high eccentricity , but these are orbits for which the `` gg '' expressions are known to lack some terms . once the `` gg
'' expressions are further improved along the lines discussed in @xcite , it is likely that the `` gg '' fluxes will always be closer to tb results .
typically , the fluxes computed from the kludge waveforms underestimate the actual energy and angular momentum lost by the orbit , and also underestimate the `` gg '' fluxes .
this suggests that inspiral waveforms that combine the two prescriptions will carry too little energy to infinity for the evolution seen in the orbit ( or equivalently , the inspiral will proceed too quickly for the energy being carried by the gws ) . in @xcite ,
a similar comparison was performed for inspirals into schwarzschild black holes and they found that the kludge gws contained _ too much _ energy , in contrast to this result .
this difference arises primarily because , in that paper , inspirals were evolved using the `` ryan '' fluxes , rather than the more up to date `` gg '' fluxes .
the signal to noise ratio squared of a monochromatic source contributed when the frequency is in the range @xmath222 is proportional to @xmath223 .
the kludge waveforms get @xmath224 largely correct ( otherwise the high overlaps would not be maintained ) , but the error in @xmath215 , arising from an amplitude discrepancy , will lead to an error in snrs computed using kludge inspiral waveforms . since @xmath215 is in general an underestimate , the kludge snrs will probably be underestimates of the true snrs .
the discrepancy in @xmath215 is at most a few tens of percent .
a discrepancy of @xmath225 _ throughout _ the inspiral would lead to an error in the snr of only @xmath226
. however , the actual error will be much less , since for the majority of the inspiral the kludge @xmath215 is much closer to the true value , and , as we saw earlier , most of the snr comes from the early inspiral .
the kludge waveforms are thus accurate enough to be used for snr calculations to estimate astrophysical event rates @xcite , in which the astrophysical uncertainties far outweigh the waveform uncertainties .
however , we must emphasize that amplitude discrepancies of this sort do not limit the applicability of the kludge waveforms for data analysis , provided the overlaps with true waveforms are high .
if the kludge waveforms were used for parameter extraction , such discrepancies would lead to an error in the distance estimate only . in practice , we are likely to use kludge waveforms for _ detection _ rather than source characterization , and for that the waveform amplitude is irrelevant .
table [ tab4 ] tabulates data for `` parabolic '' orbits , @xmath227 .
the gw emission from such orbits is important for the capture problem , i.e. , the mechanism by which compact objects are put onto emri orbits .
gravitational radiation from such orbits has been studied in the past using the teukolsky formalism @xcite,@xcite , but tabulated data has only appeared in a recent time - domain analysis by martel @xcite . those results are claimed to be accurate at the level of @xmath228 , and therefore can be treated as exact for our purposes .
table [ tab4 ] contains the relevant data together with our kludge results . for a periastron
@xmath229 the error in using the kludge fluxes is @xmath230 and rapidly decays ( grows ) as we move to larger ( smaller ) periapse . in this table
we have also included some data for non - equatorial parabolic orbits , but for this class of orbits there are as yet no available tb results .
our experience with bound generic orbits suggests it is likely these data will have accuracies of the same order as the equatorial orbits with a similar periastron . in fig .
[ parfig ] , we plot @xmath231 as a function of @xmath41 and @xmath35 ( left and right panels , respectively ) for parabolic orbits with @xmath232 and @xmath233 .
all curves are computed using the quadrupole waveform flux ( [ quad ] ) and are terminated at the point where the orbit plunges .
[ parfig ] has astrophysical significance as it displays ( within the accuracy of the present calculation ) the amount of energy lost in a single parabolic ( or @xmath4 ) encounter with a smbh , for a variety of inclinations and for a range of periastra that are appropriate to the capture problem @xcite , @xcite .
this energy loss data can be used to estimate the capture rate of compact objects by a single massive black hole .
we note that it is clear from table [ tab4 ] that the `` gg '' results perform quite poorly for parabolic orbits .
this is because the `` gg '' results are built on a small eccentricity expansion , which is no longer valid when @xmath227 . at present
the kludge fluxes are closer to tb results than the `` gg '' fluxes where comparisons can be made , and most likely are better approximations for the other orbits as well .
the performace of the `` gg '' fluxes for highly eccentric orbits is now being improved , and it is likely that the next generation prescription will be able to approximate tb fluxes accurately for orbits of all eccentricities .
in the preceding sections , we have shown that kludge waveforms do very well at reproducing the gw emission from geodesic orbits , and , if coupled to an accurate prescription for the radiation fluxes , can very accurately reproduce the gw emission from inspirals .
however , the radiation fluxes are not the whole story , as they represent only the dissipative piece of the self - force . in true inspirals
, the self - force also has a `` conservative '' component .
this does not lead to evolution of the orbital parameters and hence can not be determined from the flux of energy , angular momentum and carter constant , but does lead to a change in the phasing of the gravitational waveform .
there has been some debate recently about the importance of including the conservative pieces of the self - force when constructing templates for emri source detection @xcite .
it is quite possible that these corrections will be needed for detection , at least for some range of source parameters ; these terms will certainly be needed for parameter determination .
we aim to demonstrate here how easily conservative terms can be included in the kludge waveform prescription .
the conservative part of the self - force will include an oscillatory component , which averages to zero , and a secular piece that leads to accumulation of a phase error over time .
the net effect of the secular piece is that , for a given set of orbital parameters ( @xmath44 , @xmath43 and @xmath62 ) , the average frequencies of the motion in the @xmath47 , @xmath48 and @xmath56 directions will differ from the expected values for a kerr geodesic .
the conservative self - force might also affect the harmonic structure of the gws , but the dominant contribution will come from orbital dephasing . in the spirit of the kludge , we will now try to account for this part of the effect .
the effect of the conservative self - force can be characterized by the changes in the @xmath47 , @xmath48 and @xmath56 frequencies , averaged over many orbits .
we denote these frequency changes by @xmath234 , @xmath235 and @xmath236 .
conservative effects can then be included in the kludge orbital determination simply by amending the evolution equations for the angular variables @xmath45 , @xmath46 and @xmath56 given earlier ( eqs .
( [ dpsidt ] ) , ( [ dchidt ] ) , and ( [ geoeqphi ] ) respectively ) : @xmath237 the subscript `` geo '' denotes expressions for the phase derivatives in geodesic orbits given earlier . in general , the frequency shifts will depend on the orbital parameters , and for an inspiralling source we assume that these are evaluated for the instantaneous orbital parameters , as discussed in section [ insptraj ] . to evaluate these frequency shifts in the framework of black hole perturbation theory
requires self - force calculations @xcite which are not yet fully developed , but should be on the timescale of a few years .
once the self - force corrections are known , it should be possible to use them to compute the effective frequency shifts in the framework we use here , i.e. , objects moving on geodesics of the kerr spacetime , the parameters of which evolve with time . in the meantime , conservative corrections are already known in the post - newtonian framework up to 3.5pn order @xcite .
these can be used to compute leading order conservative corrections to include in the kludge .
we outline here a general method for such a calculation , illustrated by the simplest case of a 1pn conservative correction to circular orbits in the schwarzschild spacetime .
the generalisation to 3.5pn corrections for arbitrary orbits in kerr will be presented in a future paper .
the greatest difficulty in such a computation is to identify coordinates between the post - newtonian and perturbative formalism .
the former calculations tend to be carried out in harmonic coordinates while the natural coordinate system to use for the latter are black - hole centered coordinates , like the boyer - lindquist coordinate system .
the best approach is to use asymptotic observables , specifically the orbital frequency , perihelion precession frequency , orbital plane precession frequency and their first time derivatives , to identify coordinates between the two formalisms and find the missing conservative pieces .
the power of this approach is that it corrects the kludge in a physically meaningful way ; however , this comes at the cost of an effective @xmath238 redefinition of the orbital parameters , where @xmath239 . for extreme mass ratios , @xmath240
, this is less than the expected error in parameter determination @xcite .
the post - newtonian model has two particles moving on orbits ( expanded to the stated post - newtonian order ) and evolving via post - newtonian radiation reaction expressions .
these orbits include conservative corrections to geodesic motion .
for the case of circular orbits in schwarzschild we find at 1pn and with mass ratio corrections at linear order only that the orbital frequency and its time derivative are given by @xcite @xmath241 in the above , we use @xmath26 to denote the mass of the primary ( not the total mass as is conventional in the post - newtonian approach ) , and @xmath242 to denote the post - newtonian orbital semi - major axis . in the kludge ,
the orbit is a geodesic of the kerr spacetime , and the orbit is evolved according to the prescription in @xcite . for circular equatorial orbits in schwarzschild
we have ( at 1pn order ) @xmath243 here , @xmath47 denotes the schwarzschild radial coordinate of the orbit .
if we write @xmath244 and substitute into expression ( [ klom ] ) or ( [ klomdot ] ) , keeping corrections at 1pn order and linear in the mass ratio , we can compare to equations ( [ pnom ] ) or ( [ pnomdot ] ) and solve for @xmath245 , @xmath246 and @xmath247 .
the comparison between ( [ klom ] ) and ( [ pnom ] ) gives a different result for @xmath246 and @xmath247 than the comparison between ( [ klomdot ] ) and ( [ pnomdot ] ) , since the kludge prescription ignores conservative effects . we therefore write @xmath248 and solve simultaneously for @xmath245 , @xmath246 , @xmath247 , @xmath249 and @xmath250 [ bearing in mind that ( [ klrdot ] ) not ( [ klomdot ] ) is the fundamental quantity to examine for computing inspiral in the kludge ] .
we find that the required parameters are @xmath251 we note that the correction to @xmath252 scales as @xmath253 , compared to the scaling of @xmath254 , which is @xmath255 . in the weak - field ,
@xmath256 , the conservative correction will be much more significant than the radiative correction .
this is in keeping with what pound et al .
@xcite found for a particle moving under an electromagnetic self - force .
extension to generic orbits is straightforward . in that case
, we have two extra observables , the perihelion precession frequency and the rate of precession of the orbital plane , and two extra coordinate dependent orbital parameters , the eccentricity and orbital inclination .
identifying the rate of change of the orbital frequency , the perihelion precession frequency and the orbital plane precession frequency between the post - newtonian and kludge approaches gives us a relation between the two sets of coordinates .
comparing the values of the frequencies then tells us which @xmath236 , @xmath257 and @xmath258 must be added into the kludge . in all of this
we have chosen implicitly to identify the two masses , @xmath26 and @xmath16 , and the spin , @xmath60 , between the post - newtonian and kludge models . to our minds
, it makes more sense to hold these parameters fixed between the two models , at the cost of modifying the definitions of the orbital elements @xmath259 .
once self - force data is available , it will be possible to obtain a consistent solution for the conservative correction in boyer - lindquist like coordinates by using fits to the self - force results . in using equation ( [ klrdot ] ) ,
we have implicitly assumed the relationship between energy / angular momentum and the radial coordinate is unchanged by the self - force .
however , in self - force calculations , the energy and angular momentum of circular orbits at a given radius do change as a result of conservative effects .
this is merely a manifestation of the @xmath238 redefinition of @xmath47 implicit in this approach that we mentioned earlier . when comparing to self - force calculations , it will be preferable to include this redefinition of energy as well as the conservative corrections to @xmath252 so that @xmath47 maintains it s meaning . ultimately , both approaches are equivalent .
another way to obtain @xmath249 and @xmath250 for equation ( [ klpluscons ] ) is by expanding @xmath260 as a function of @xmath252 in both the pn and kludge approaches .
matching various orders in this expansion gives @xmath249 and @xmath250 directly without having to simultaneously solve for the coordinate transformation .
the function @xmath261 is a gw observable , so this approach ensures that the kludge will have the correct leading order form for this observable .
however , this is a series truncated at a given pn order , so it will be important to assess how significant the omitted terms can be .
it is likely that using the present kludge , augmented with conservative corrections up to 3.5pn order computed as outlined above , will do well enough at reproducing the phasing of true inspiral waveforms . as discussed in @xcite , it is likely that the conservative terms will be most important in the weak - field regime , where the post - newtonian results are valid . in the regime where the post - newtonian corrections cease to be valid
, the contribution of conservative corrections to the phase evolution may be much less critical .
this will be investigated in future work .
if the pn expansion is not sufficient , the kludge can be augmented using fits to the results of self - force calculations as described .
in this paper we have provided a simple , `` easy - to - use '' , prescription for approximating the gravitational waveforms generated by test - bodies inspiralling into kerr black holes .
these `` kludge '' waveforms are constructed by combining familiar flat - spacetime wave equations with a true geodesic trajectory in the kerr spacetime . despite its formal inconsistency this hybrid approximation results in remarkably accurate waveforms , which we have established by comparison to rigorous teukolsky - based perturbative waveforms .
we find an impressive overlap between the kludge and tb waveforms for particles on geodesic orbits ( i.e. , ignoring radiation - reaction ) .
this overlap is @xmath262 over a significant portion of the relevant orbital parameter space
. significant degradation ( overlap reduced to @xmath155 ) occurs for strong field orbits around rapidly spinning black holes . for such cases ,
the contribution of the radiation backscattered from the background spacetime is sizable ; this effect is not included in the present formulation of the kludge waveforms . as a rough ( but reliable ) rule of thumb
, we have found that kludge waveforms work well for kerr orbits with periaspe distance @xmath263 , irrespective of the black hole spin .
we have experimented with three different types of kludge waveforms generated using three different solutions of the flat - spacetime gravitational wave - equation : quadrupole , quadrupole - octupole and `` press '' waveforms @xcite .
the latter choice includes contributions from all multipole moments of the orbiting body .
we have concluded that for all practical purposes the quadrupole - octupole waveforms are the optimal choice , nearly as accurate as the press waveforms while much quicker and easier to generate . within the adiabatic approximation ,
we have applied our method to calculate full inspiral kludge waveforms taking into account orbital evolution .
this realistic scenario requires an additional `` kludge '' for describing the orbital evolution itself .
our earlier work @xcite,@xcite provides such a scheme : exact geodesic orbital dynamics coupled with approximate pn - based expressions for the fluxes .
the resulting waveforms still have very high overlaps with available teukolsky - based inspirals for circular - inclined orbits .
the overlap is @xmath264 even when the inspiral terminates in the strong field region .
this result indicates the kludge will be a very useful tool for generating inspiral waveforms .
however , as yet teukolsky - based inspirals are available only for circular - inclined orbits . only when comparisons for generic inspiral orbits are possible will we be able to firmly identify the range of applicability of these kludge waveforms . we have also used the kludge waveforms to estimate the fluxes of energy and angular momentum carried away in gws from geodesic orbits .
we have found that such kludge fluxes in general are not as accurate for evolving inspirals as the post - newtonian based expressions described in @xcite .
the kludge fluxes do provide us with an estimate of the error in using kludge waveforms for snr calculations , and our results suggest the kludge waveforms will tend to underestimate the true snrs if anything . the area where these kludge waveforms will find most use is in the development of emri data analysis for lisa .
the combination of accuracy and simplicity of generation has already made these waveforms invaluable tools for the study of data analysis issues .
it seems quite plausible that the waveforms may also play a role in the final search of the lisa data .
one use could be for estimation of the waveform parameters as the first stage in a hierarchical search .
the high faithfulness of these waveforms suggest that they may be able to set fairly tight bounds on the parameters of the emitting system
. this will be extremely useful input for the second stage of the search where the system parameters will be refined using more accurate waveforms .
however , as we discussed before , the kludge waveforms at present do not include some important physical features that we expect in true inspirals .
only when accurate teukolsky or self - force based waveforms are available for generic inspiral orbits will we be able to fully quantify the range of validity and level of accuracy of the kludge waveforms .
time - frequency searches @xcite could also be used to detect emris in the first stage of a hierarchical search .
these can typically detect emri events at about half the distance of the semi - coherent search @xcite , without the use of templates . a template with overlap of @xmath265 with the true signal can detect that signal at about half the distance of the correct template .
the nk waveforms can easily achieve overlaps of @xmath265 through most of parameter space , particularly given the @xmath266 week integration times needed for the semi - coherent approach .
it is clear therefore that a first stage semi - coherent search using kludge waveforms would have a greater reach than time - frequency methods , and would put tighter constraints on the parameters .
however , this would come at a much greater computational cost .
the final lisa search will undoubtedly employ both methods at various stages in the analysis .
a further possible application of kludge waveforms is to the study of non - kerr emris .
it is hoped that lisa observations will allow `` spacetime - mapping '' of black holes @xcite , @xcite , and thereby test the no - hair theorem . to carry out such tests
quantitatively will require waveform templates , which incorporate the deviation from kerr in the spacetime structure as a set of suitable parameters ( e.g. , multipole moments @xcite ) .
the development of rigorous non - kerr emri waveforms is a very difficult task , since generic spacetimes lack some symmetries that allowed the formulation of the teukolsky framework for the kerr spacetime .
one way to make progress is to construct kludged waveforms in non - kerr spacetimes along the lines outlined in this paper .
our results for the kerr spacetime suggest that such kludged waveforms may well be sufficiently accurate for qualitative , and even quantitative studies of non - kerr emris .
all that is required is the integration of geodesic equations in the non - kerr spacetime and thus kludge waveforms provide a computationally quick and easy tool to study `` bumpy '' and `` quasi - kerr '' spacetime mapping @xcite .
the kludged waveforms / inspirals presented in this and a companion paper @xcite could be considered as `` second generation '' , an improvement of the original simple version of quadrupole waveforms @xcite and inspirals @xcite . certainly , there is space for further improvement .
the inclusion of conservative self - force effects on the inspiral is an obvious next step , and we have already discussed here how this can be achieved by calculating the relevant orbital frequency shifts for circular schwarzschild orbits .
we can also further improve waveform generation by including the back - scattering effect arising from the propagation of the gws in a curved spacetime . in summary ,
given the present level of performance of our kludge waveforms and inspirals and their prospects of improvement , we feel that they should remain valuable tools for lisa source modelling for the coming years .
we thank kip thorne for initially pointing us to the paper by press , steve drasco for providing teukolsky based waveforms from generic orbits and b.s .
sathyaprakash for useful discussions .
the work of jrg was supported in part by nasa grants nag5 - 12834 and nag5 - 10707 and by st.catharines college , cambridge .
kg acknowledges support from pparc grant ppa / g / s/2002/00038 .
work of sb was partially supported by pparc grant pp / b500731 .
work of hf was supported in parts by nasa grants nag5 - 12834 , nng04gk98 g and by nsf grants phy-0099568 and phy-0601459 .
the work of sah was supported by nasa grants nag5 - 12906 and nng05g105 g , nsf grant phy-0244424 and career grant phy-0449884 .
sah also gratefully acknowledges support from mit s class of 1956 career development fund .
.[o - lap_equat ] numerical data for overlaps between tb and kludge waveforms equatorial - eccentric kerr orbits .
data is not shown with wd confusion noise for the ten orbits in the very strong field regime . in this
regime , none of the kludge waveforms reproduce the tb waveforms very well , and this is compounded when the dominant harmonics are suppressed by white dwarf confusion noise .
the overlaps are uniformly poor and uninformative , so we do not include them . [ cols="^,^,^,^,^,^,^,^,^,^,^,^ " , ] blanchet , l. , `` gravitational radiation from post - newtonian sources and inspiralling compact binaries '' , _ living rev . relativity _ 5 , ( 2002 ) , 3 .
[ online article ] : cited on 4/4/2006 , http://www.livingreviews.org/lrr-2002-3 | one of the most exciting potential sources of gravitational waves for low - frequency , space - based gravitational wave ( gw ) detectors such as the proposed laser interferometer space antenna ( lisa ) is the inspiral of compact objects into massive black holes in the centers of galaxies .
the detection of waves from such `` extreme mass ratio inspiral '' systems ( emris ) and extraction of information from those waves require template waveforms .
the systems extreme mass ratio means that their waveforms can be determined accurately using black hole perturbation theory .
such calculations are computationally very expensive .
there is a pressing need for families of approximate waveforms that may be generated cheaply and quickly but which still capture the main features of true waveforms . in this paper
, we introduce a family of such `` kludge '' waveforms and describe ways to generate them .
different kinds of `` kludges '' have already been used to scope out data analysis issues for lisa .
the models we study here are based on computing a particle s inspiral trajectory in boyer - lindquist coordinates , and subsequent identification of these coordinates with flat space spherical polar coordinates . a gravitational waveform
may then be computed from the multipole moments of the trajectory in these coordinates , using well known solutions of the linearised gravitational perturbation equations in flat spacetime .
we compute waveforms using a standard slow - motion quadrupole formula , a quadrupole / octupole formula , and a fast - motion , weak - field formula originally developed by press .
we assess these approximations by comparing to accurate waveforms obtained by solving the teukolsky equation in the adiabatic limit ( neglecting gw backreaction ) .
we find that the kludge waveforms do extremely well at approximating the true gravitational waveform , having overlaps with the teukolsky waveforms of @xmath0 or higher over most of the parameter space for which comparisons can currently be made . indeed
, we find these kludges to be of such high quality ( despite their ease of calculation ) that it is possible they may play some role in the final search of lisa data for emris . |
consider a polygon @xmath4 , whose @xmath5 and @xmath6 coordinates are given by the vectors @xmath7 and @xmath8 , respectively .
consider a transformation on @xmath4 which creates a new polygon @xmath9 by taking the midpoints of the original , connecting them , and then normalizing the resulting vertex vectors to unit length .
it has been shown that iteration of this process produces a sequence of polygons @xmath10 converging to an ellipse oriented at @xmath11 degrees @xcite . our goal is to relax the condition of taking only the midpoint , and allow the point of division @xmath12 to vary inside the open unit interval .
indeed , there is an elegant and surprising invariance in the orientation of our ellipse that hints at a deeper underlying structure for this type of problem .
we note that this paper will make absolutely no assumption on our initial polygon .
we do not require convexity or any other special condition that makes our polygon well - behaved .
this is truly an example of order out of pure chaos . indeed
, there are numerous papers written on the subject of iterative transformations of random polygons , in which limiting behavior is of interest .
many of these papers hint at an underlying invariance and structure of these types of iterations ( see @xcite , @xcite , @xcite , and @xcite ) .
in proving the case for @xmath3 a matrix @xmath13 was introduced such that if @xmath14 is the x coordinates of our vertex vectors after the kth iteration , then @xmath15 .
we then wish to study the sequence @xmath16 , where @xmath17 denotes the polygon @xmath18 with vertex coordinates given by @xmath5 and @xmath6 , respectively .
since the polygon is entirely determined by these two vectors , we shall only study how each vertex vector changes with respect to our transformation . in this way
, we find that our question of convergence becomes a question of the properties of iterating @xmath13 .
following in these tracks , we introduce a more general version that applies to any @xmath12 .
let @xmath19 with @xmath13 defined as so , we can turn to the analysis of its properties . without normalizing after every iteration , it can be shown that iteration of @xmath13 will produce a sequence that eventually converges to the centroid of our original polygon . the sequence of vertex vectors @xmath20 and @xmath21 created by iteration of @xmath13 has the same centroid for every @xmath22 .
letting @xmath23 represent an @xmath24 matrix of 1 s , we note that the centroid of the @xmath5 coordinates of our polygon is just @xmath25 , and the centroid of the @xmath6 coordinates is defined analogously .
we then see : @xmath26 from this , we deduce that if our sequence converges to a point , that point is the centroid of our initial polygon , and that every polygon in the sequence has the same centroid .
intuitively , it is clear that @xmath27 \subset \big [ \min \ { \vec{x}^{(k ) } \ } , \max \ { \vec{x}^{(k ) } \ } \big]\ ] ] where @xmath28 denotes the largest component of the vector @xmath29 ( and the minimum defined analogously ) .
if we had a contraction after every iteration , it is obvious that the sequence of nested intervals must converge to a point . not
every case will induce a contraction , though , with the obvious example being a square .
however , since our polygon must have a finite number of vertices , after a sufficiently large amount of iterations , our transformation intuitively should induce a contraction . indeed
, we will make this intuition more rigorous in section 5 . by introducing the upshift matrix @xmath30 @xcite , which is just the identity matrix with all of its entries shifted up once
, we see that @xmath31 in this way , we see that @xmath13 shares the same eigenvectors as @xmath30 but with different eigenvalues .
since @xmath30 is unitary , its eigenvectors will be orthogonal . by normalizing we create an orthonormal set of eigenvectors where the eigenvalues of @xmath30 are @xmath1th roots of unity .
allow that @xmath32 and @xmath33 it is easy to see that @xmath34 . in this way , we can also see that @xmath35 so that the eigenvectors of @xmath30 are also eigenvectors of @xmath13 .
the eigenvalues of @xmath13 lie on the circle of radius @xmath12 centered at @xmath36 on the complex plane . .
from top to bottom the initial polygons have @xmath37 , @xmath38 , and @xmath39 vertices , respectively .
we see that the transformation does seems to tend to the centroid , and smaller @xmath1 seem to converge faster . ] . from top
to bottom the initial polygons have @xmath37 , @xmath38 , and @xmath39 vertices , respectively .
we see that the transformation does seems to tend to the centroid , and smaller @xmath1 seem to converge faster . ] . from top
to bottom the initial polygons have @xmath37 , @xmath38 , and @xmath39 vertices , respectively .
we see that the transformation does seems to tend to the centroid , and smaller @xmath1 seem to converge faster . ]
from @xcite we see that our analysis leads in the direction of repeated iteration of @xmath13 onto a unit vector with centroid zero in the span of our eigenvectors .
we can assume that our unit vector has centroid zero because we can always just redefine the centroid as the origin of our coordinate system .
this seemingly innocent assumption simplifies our analysis because it in fact forces our eigenvectors to be a dependent set , and the eigenvector corresponding to the eigenvalue @xmath41 turns out to be redundant .
it can then be seen that our vectors will becomes increasingly rich in the second largest eigenvalue and its complex conjugate .
the rate at which our other eigenvectors converge can be measured by a _ damping factor _
@xmath40 defined as the magnitude of the ratio of the third largest eigenvalue to the second largest .
the reason we consider this is based off of an approximation method known as the _ power method_. consider : @xmath42 this definition makes the most sense if we reorder our eigenvalues in such a way that @xmath43 we now seek to find an explicit damping factor .
@xmath44 ^{\frac{1}{2}}\ ] ] with some basic algebra this can be simplified to : @xmath45^{\frac{1}{2}}\ ] ] with , we deduce the following : for any @xmath0 , @xmath40 attains a minimum at @xmath3 .
differentiate @xmath40 with respect to @xmath12 .
@xmath46 we have : @xmath47 after some simplification , we are left with : @xmath48 @xmath49 it is interesting to note that the above proof does not specify any value for @xmath1 .
indeed , this condition holds for any arbitrary @xmath1 . with the damping factor
now specified , it is clear that upon iteration the damping factor will eventually converge to 0 .
heuristically , the damping factor makes sense because as our sequence tends to infinity , the largest eigenvalues will of course dictate the overall behavior of the sequence . by the above lemma we see that the midpoint yields the fastest possible rate of convergence .
we now intend to specify what this sequence is actually converging to . .
from top to bottom we have chosen @xmath50 , @xmath51 , and @xmath52 .
it can be seen that as @xmath12 tends closer to @xmath53 , the rate of convergence is faster . ]
we introduce an invariant subspace under our operator @xmath13 which is defined as @xmath55 where @xmath56 is a reordering of our eigenvectors in such a way that @xmath57 . with the material of the previous section , we can introduce the undamped ( unit normed ) vector @xmath58 where @xmath59 are arbitrary constants . in this way
, we define @xmath60 where @xmath61 is a centroid zero unit vector in @xmath62 . with this
, we see that @xmath63 since @xmath64 for all @xmath0 , it is clear that our vector must eventually converge to @xmath54 .
we now show the underlying structure of @xmath54 by the introduction of a new set of coordinates .
let @xmath65 in matrix notation we can define @xmath66 as the matrix with @xmath67 in the jth row .
we define @xmath68 analogously .
it can be shown @xcite that this is indeed an orthonormal set upon normalizing and defining @xmath69 and @xmath70 we now seek to examine this set @xmath71 under iteration of our matrix @xmath13 , where it is clear that this set is indeed an orthonormal basis for @xmath72 .
define @xmath73 where @xmath74 is arbitrary . noting that @xmath75
, we have : @xmath76 @xmath77 @xmath78 similarly , we can find that @xmath79 in matrix notation , we have : @xmath80 @xmath81 in order to study the iteration of @xmath13 on these two basis elements , it is crucial to first introduce suitable notation .
let @xmath82 and @xmath83 now let @xmath84 and @xmath85 setting @xmath86 , we can now put the above equation in a much more succinct form .
we have : @xmath87 @xmath88 in this form our two expressions resemble a recurrence relation , which motivates us to look for a closed form of @xmath89 and @xmath90 .
we see that it in fact satisfies a strikingly symmetric relationship .
let @xmath91 and @xmath92 be defined as above .
let @xmath93 , where @xmath94 denotes the imaginary unit .
we then have : @xmath95 @xmath96 by induction .
base case is trivial .
we proceed inductively : @xmath97 @xmath98 collecting terms , we see the coefficient of @xmath99 is : @xmath100 @xmath101 @xmath102 where the angle sum formula was used for the last step .
likewise we find the coefficient for @xmath103 as : @xmath104 @xmath105 as desired .
the case for @xmath106 is nearly identical .
if we rewrite the results given above a little differently , the elegance and symmetry of the two expressions becomes more striking .
we have : @xmath107 @xmath108
we now summarize a collection of the above results to continue below . without loss of generality
, we can assume our given polygon has centroid at the origin . under this assumption , we then know that both vectors @xmath7 and @xmath8 are orthogonal to @xmath109 , the @xmath24 vector of all 1 s .
thus : @xmath110 from here , we can express @xmath14 and @xmath111 as : @xmath112 @xmath113 where @xmath114 @xmath115 and @xmath116 , @xmath117 are arbitrary scalars .
let us now assume that @xmath118 is a given unit vector such that @xmath119 we want to examine the behavior of @xmath120 . using the results of the previous section
, we can now do this and find a rather simple closed form .
@xmath121 @xmath122 similarly , we see @xmath123 with this result we are now able to prove the following lemma . for any arbitrary point of division @xmath0 , iteration of @xmath13 onto the vertices of a given polygon will converge to the centroid of the polygon , and every polygon in the sequence will have the same centroid . from lemma 2.1
we have already shown that if our sequence converges to a point , then the hypothesis holds .
we must then show that our vertices will in fact converge to a point .
since we have already shown our vertex vectors @xmath124 , @xmath125 will become arbitrarily close to @xmath120 and @xmath126 , respectively , we only need to show convergence of the latter two vectors . from and , we then deduce : @xmath127 for @xmath0 , @xmath128 . thus , @xmath129 @xmath130 therefore @xmath131 for all @xmath0 ( similarly for @xmath132 ) , so our sequence converges to a single point , the centroid . * corollary*. the sequence of polygons @xmath16 such that @xmath3 will converge most quickly to the centroid of @xmath133 .
we note that @xmath134 is positive everywhere except its root @xmath3 . from the work of the above lemma
, we have : @xmath135 with equality if and only if @xmath3 .
we now turn our attention back to the behavior of our normed vectors @xmath136 and @xmath137 . from and
, we see that the magnitudes of these vectors are only dependent on the magnitude @xmath138 , and so we easily deduce the normed vertex vectors : @xmath139 @xmath140 from @xcite we find that our ellipse can be completely specified by the singular value decomposition ( svd ) of the set given by : @xmath141 where @xmath142 .
we now seek to find the svd of our 2x2 matrix .
this is because , if we put a matrix @xmath143 , then the matrices @xmath144 and @xmath145 will specify the tilt of our ellipse , whose semiaxes are specified by the diagonal elements of our matrix @xmath146 . if @xmath147 then there exist matrices @xmath144 , @xmath146 , and @xmath145 such that @xmath143 , where @xmath148 @xmath149 @xmath150 it can be shown @xcite that the singular value decomposition for any 2x2 matrix is of the form @xmath151 the rest is omitted as it is not essential to understanding the properties of our ellipse .
the explicit forms are found by straightforward computation of the explicit forms of the svd of a 2x2 matrix ( see @xcite ) .
firstly , we see that by the matrices and , our ellipse will still be oriented at 45 degrees no matter what @xmath0 is chosen .
one of the properties to be analyzed can be deduced from the matrix : periodic semi - axes .
this can be seen by the explicit forms of each semi - axes of our ellipse , which are given by : @xmath152 @xmath153 , @xmath51 , and @xmath52 , respectively .
we see that each @xmath12 tends to the ellipse at @xmath11 degrees , and values closer to @xmath53 converge more quickly . ] , @xmath51 , and @xmath52 , respectively .
we see that each @xmath12 tends to the ellipse at @xmath11 degrees , and values closer to @xmath53 converge more quickly . ] , @xmath51 , and @xmath52 , respectively .
we see that each @xmath12 tends to the ellipse at @xmath11 degrees , and values closer to @xmath53 converge more quickly . ]
recall that @xmath154 @xmath140 by the periodicity of the trigonometric functions , we deduce an interesting property of the @xmath54 ellipse . for @xmath93 , where @xmath91 and @xmath92 defined as before , @xmath136 is periodic with period @xmath155 if and only if @xmath156 is a rational multiple of @xmath157 .
assume @xmath158 , where @xmath159 .
we have : @xmath160 from here we see that @xmath161 must vanish after the use of elementary trigonometric identities .
thus @xmath162 where @xmath163 , @xmath159 .
also , @xmath164 , and likewise for @xmath99 . therefore we see that this is periodic since the upshift matrix is just a reordering of our entries .
the converse is trivial .
although this is a rather idealized case , in practice it seems that there still exists some periodicity , or _ almost periodicity _
, for any point of division .
this stems from the fact that for any given argument of @xmath165 as defined , it can be approximated arbitrarily close to some rational multiple of @xmath157 .
explicitly , we have : @xmath166 where we allow that @xmath163 be any arbitrary integer .
we do not hold @xmath163 to be fixed , we allow it to vary . because of this , as @xmath22 , @xmath156 will be sufficiently close to a rational multiple of @xmath157 an arbitrary amount of times . to illustrate this more clearly ,
consider taking @xmath38 randomized vertices with @xmath167 .
then , we first start taking @xmath168 .
we now scale this by some integer such that our value is very close to an integer . in this case , it is clear that scaling by @xmath169 gets us close to the number @xmath170 , so we could expect that this is weakly periodic with a period of @xmath170 . continued scaling shows us other places where we get even closer to integer values . for instance , scaling by @xmath171 gives us a number extremely close to @xmath172 , and it is to be expected that this process can be carried out indefinitely for closer and closer approximations .
in fact , this is an illustration of _ dirichlet s approximation theorem _ , which gives us the following fact : given the number @xmath157 , there exists an infinite sequence of rational numbers such that @xmath173 of course , this is just a particular case , and dirichlet s theorem has a much more general statement .
finally , we consider our semi - axes .
as is clearly seen by our matrix , we have semi - axes that are dependent on @xmath174 .
this leads us to deduce a property of oscillating semi - axes . in practice
, however , this oscillation is very slight and almost impossible to see for larger @xmath1 , as can be predicted by the fact that our periodicity gets much larger as the number of vertices grows .
we now consider the case of taking a different point of division at each segment .
we can now use the results given so far to generalize even further .
suppose that we are given a polygon @xmath175 and instead of using a fixed point of division @xmath12 for every segment , we give an arbitrary and different point of division @xmath176 to the @xmath94th segment of our polygon .
this case is inherently similar , and can in fact be reduced to the properties of the sequences of polygons @xmath177 , where @xmath178 denotes the sequence of polygons produced by our iteration when using the fixed point of division @xmath176 for each segment . in this way , it is easy to see that we can express our random point of division polygon , denoted @xmath179 as : @xmath180 with this form , we see that iteration of @xmath13 on the vertex vectors of @xmath179 without normalization will will also converge to the centroid of the original polygon , since we have already proved that every polygon in the intersection has to converge to the centroid .
we now consider the case of @xmath181 , which is the normalized iteration .
it can be easily shown that our periodicity will exist if and only if @xmath156 is a rational multiple of @xmath157 and that our period will become @xmath182 $ ] , where @xmath183 is the period of the ellipse produced by @xmath176 .
i would like the thank jimmy dillies for introducing me to this problem , as well as scott kersey for aiding me in creating the matlab code to model this problem .
jiu ding , l richard hitt , xin - min zhang , _ markov chains and dynamic geometry of polygons _ ,
linear algebra and its applications , volume 367 , 1 july 2003 , pages 255 - 270 , issn 0024 - 3795 , http://dx.doi.org/10.1016/s0024-3795(02)00634-1 . | this paper generalizes the result of elmachtoub et al to any weighted barycenter , where a transformation is considered which takes an arbitrary point of division @xmath0 of the segments of a polygon with @xmath1 vertices .
we then consider connecting these new points to form another polygon , and iterate this process . after considering properties of our generalized transformation matrix , a surprisingly elegant interplay of elementary complex analysis and linear algebra
is used to find a closed form for our iterative process .
we then specify the new limiting ellipse , @xmath2 , which has oscillating semi - axes . along the way we find that the case for @xmath3 enjoys some special optimality conditions , and periodicity of the ellipse @xmath2 is analyzed as well . to conclude ,
an even more generalized case is considered : taking a different point of division for every segment of our polygon @xmath4 . |
in later years , use of non - ultralocal actions which preserve chiral symmetry on the lattice has become more viable .
the two actions which are being used most frequently are overlap fermions @xcite based on the wilson fermion action and domain - wall fermions @xcite .
overlap fermions are notoriously difficult to study , both numerically and analytically .
many recent promising investigations involving simulations with overlap fermions have appeared ; see , e.g. , refs .
@xcite . regarding analytical computations ,
the only ones performed thus far have been either up to 1 loop , such as refs .
@xcite , or vacuum diagrams at higher loops @xcite . _
the present work is the first one involving non - vacuum diagrams beyond the 1-loop level .
_ we compute the 2-loop renormalization @xmath6 of the bare lattice coupling constant @xmath0 in the presence of overlap fermions .
we relate @xmath0 to the renormalized coupling constant @xmath2 as defined in the @xmath11 scheme at a scale @xmath12 ; at large momenta , these quantities are related as follows @xmath13 ( @xmath14 lattice spacing ) .
the 1-loop coefficient @xmath15 has been known for a long time ; several evaluations of @xmath16 have also appeared in the past @xmath1710 years , either in the absence of fermions @xcite , or using the wilson @xcite or clover @xcite fermionic actions .
knowledge of @xmath16 , along with the 3-loop @xmath1-renormalized @xmath9-function @xcite allows us to derive the 3-loop bare lattice @xmath9-function , which dictates the dependence of lattice spacing on @xmath0 .
in particular , it provides a correction to the standard 2-loop asymptotic scaling formula defining @xmath18 .
ongoing efforts to estimate the running coupling from the lattice @xcite have relied on a mixture of perturbative and non - perturbative investigations . as a particular example , relating @xmath19 to @xmath20 ( sf : schrdinger functional scheme , as advocated by the alpha collaboration ) , entails an intermediate passage through the bare coupling and the conversion from @xmath19 to @xmath21 is carried out perturbatively .
the paper is organized as follows : some theoretical background and the methodology of the necessary perturbative calculations are given in section ii .
section iii regards the overlap action where the derivation of the vertices with up to 4 gluons is explained .
we also provide the expressions for these vertices ( the 4-gluon vertex is given in appendix a ) .
details on our computation , numerical results and plots of @xmath6 coefficients and the @xmath9-function can be found in section iv . finally , in section v we give the summary and our conclusions . the present study , being the first of its kind in calculating 2-loop diagrams with overlap vertices and external momentum dependence , had a number of obstacles to overcome .
one first complication is the size of the algebraic form of the feynman vertices ; as an example , the vertex with 4 gluons and a fermion - antifermion pair contains @xmath17724,000 terms when expanded . upon contraction
these vertices lead to huge expressions ( many millions of terms ) ; this places severe requirements both on the efficiency of the computer algorithms which we must design to manipulate such expressions automatically and on the necessary computer ram . numerical integration of feynman diagrams over loop momentum variables is performed on a range of lattices , with finite size @xmath22 , and subsequent extrapolation to @xmath23 .
as it turned out , larger @xmath22 were required for an accurate extrapolation in the present case , compared to ultra - local actions .
in addition , since the results depend nontrivially on the parameter @xmath7 of the overlap action , numerical evaluation must be performed for a sufficiently wide set of values of @xmath7 , with an almost proportionate increase in cpu time .
extreme values of @xmath7 ( @xmath24 ) show unstable numerical behaviour , which is attributable to the spurious poles of the fermion propagator at these choices ; this forces us to even larger @xmath22 .
a consequence of all complications noted above is an extended use of cpu time : our numerical integration codes , which ran on a 32-node cluster of dual cpu pentium iv processors , required a total of @xmath1750 years of cpu time .
the definition and value of the renormalized coupling constant @xmath25 depends on the renormalization scheme ( parameterized by a scale @xmath26 ) , and this dependence is given by the renormalized @xmath9-function @xmath27 we will adopt the @xmath1 renormalization scheme throughout this work , and we will denote the renormalized coupling constant @xmath2 simply by @xmath25 . for the lattice regularization
a bare @xmath9-function is defined as @xmath28 where @xmath12 the renormalization scale and @xmath25 @xmath29 the renormalized ( bare ) coupling constant .
it is well known that in the asymptotic limit for qcd ( @xmath30 ) , one can write the expansion of the @xmath9-function in powers of @xmath0 , that is @xmath31 the coefficients @xmath32 are universal constants ( regularization independent ) given by @xmath33 \label{univ_coeff}\end{aligned}\ ] ] where @xmath34 is the number of colors ( flavors ) .
+ on the contrary , @xmath35 ( @xmath36 ) depends on the regulator ; it must be determined perturbatively . in the present work
we calculate the coefficient @xmath37 using the @xmath38 fermionic action and wilson gluons .
the perturbative expansion of the renormalized @xmath9-function is @xmath39 @xmath40 and @xmath41 can be related using the renormalization function @xmath6 , defined through @xmath42 , that is could be denoted : @xmath43 to indicate its dependence on the regulator ( @xmath22 : lattice ) and on the renormalization scheme ( @xmath1 ) . ]
@xmath44 computing @xmath45 to 2 loops @xmath46 and inserting it in eq .
( [ beta1 ] ) , allows us to extract the 3-loop coefficient @xmath47 .
the quantities @xmath48 and @xmath49 have been known in the literature for quite some time @xcite ; @xmath50 and @xmath51 are the same as those of the bare @xmath9-function , eq .
( [ univ_coeff ] ) , and @xmath52 in the @xmath53 scheme is @xmath54 \label{b2ms}\ ] ] the constant @xmath49 is related to the ratio of the @xmath55 parameters associated with the particular lattice regularization and the @xmath56 renormalization scheme @xmath57 for overlap fermions the exact form of @xmath49 appears in ref .
@xcite @xmath58 \label{l0}\ ] ] where @xmath59 is the convergent part of the 1-loop fermionic contribution ( denoted by @xmath60 in ref .
@xcite ) , presented in table i. eq .
( [ beta1 ] ) is valid order by order in perturbation theory and expanding it in powers of @xmath61 the first nontrivial relation is @xmath62 thus , the evaluation of @xmath37 requires only the determination of the 2-loop quantity @xmath63 .
a direct outcome of our calculation is the 2-loop corrected asymptotic scaling relation between @xmath64 and @xmath0 @xmath65 , \qquad q = { b_1 ^ 2-b_0b_2^l\over 2b_0 ^ 3 } \label{asympre}\ ] ] where all quantities in the correction term @xmath66 , except @xmath47 , are known .
the most convenient and economical way to proceed with calculating @xmath67 is to use the background field technique @xcite , in which the following relation is valid @xmath68 where @xmath69 , defined as : @xmath70 is the background field renormalization function ( @xmath71 ( @xmath72 ) : bare ( renormalized ) background field ) . in the lattice version of the background field technique
, the link variable takes the form @xmath73 ( @xmath74 : quantum field , @xmath75 : background field ) . in this framework , instead of @xmath67 , it suffices to compute @xmath76 , with no need to evaluate any 3-point functions . for this purpose
, we consider the background field one - particle irreducible ( 1pi ) 2-point function , both in the continuum ( dimensional regularization , @xmath1 subtraction ) : @xmath77 and on the lattice : @xmath78 . in the notation of ref .
@xcite , these 2-point functions can be expressed in terms of scalar functions @xmath79
@xmath80/g_0 ^ 2 , \qquad\qquad\qquad\quad \nu(p ) = g_0^{2 } \nu^{(1)}(p)+g_0^{4 } \nu^{(2)}(p)+ ... \label{2pt - functiona}\end{aligned}\ ] ] ( @xmath81 ) . there follows @xmath82 the gauge parameter @xmath83 must also be renormalized ( up to 1 loop ) , in order to compare lattice and continuum results @xmath84 ( @xmath85 : renormalization function of the quantum field ) .
the coefficient @xmath86 is obtained from the quantum field 1pi 2-point function in the continuum ( @xmath87 ) and on the lattice ( @xmath88 ) through @xmath89 , \quad\ , \omega_r(p ) = g^{2 } \omega_r^{(1)}(p)+{\cal o}(g^{4 } ) \label{2pt - functionq_r}\\ \nonumber \\ & & \sum_\mu \gamma^{qq}_l(p)^{ab}_{\mu\mu } = -\delta^{ab}\widehat{p}^2 \left [ 3\left ( 1 - \omega(p)\right ) + \lambda_0\right ] , \qquad\qquad\qquad\quad\ , \omega(p ) = g_0^{2}\ , \omega^{(1)}(p)+{\cal o}(g_0^{4 } ) \label{2pt - functionq}\end{aligned}\ ] ] @xmath90 in terms of the perturbative expansions eqs .
( [ 2pt - functiona_r ] ) , ( [ 2pt - functiona ] ) , ( [ 2pt - functionq_r ] ) , ( [ 2pt - functionq ] ) , @xmath45 takes the form @xmath91_{\lambda=\lambda_0 } \label{zg_expanded}\ ] ] the fermion part of @xmath92 coincides with that of @xmath93 .
similarly for the fermion part of @xmath94 and @xmath95 .
consequently , one may write @xmath96_{n_f=0 } + [ \nu^{(1 ) } - \nu_r^{(1 ) } ] - [ \nu^{(1 ) } - \nu_r^{(1)}]_{n_f=0}\end{aligned}\ ] ] since the quantities of interest are gauge invariant , we choose to work in the bare feynman gauge , @xmath97 , for convenience . in order to compute @xmath69
we need the expressions for @xmath98 .
the @xmath56 renormalized functions necessary for this calculation to 2 loops are @xcite @xmath99 + { n_f\over 16\pi^2}\left [ { 2\over 3}\ln{p^2\over \bar{\mu}^2 } - { 10\over 9}\right ] \label{nu1}\\ \nonumber \\
\omega_r^{(1)}(p,\lambda)= & & { n\over 16\pi^2 } \left [ \left(-{13\over 6}+{1\over 2\lambda}\right ) \ln{p^2\over \bar{\mu}^2 } + { 97\over 36 } + { 1\over 2\lambda } + { 1\over 4 \lambda^2}\right ] + { n_f\over 16\pi^2}\left [ { 2\over 3}\ln{p^2\over \bar{\mu}^2}- { 10\over 9}\right ] \label{omega1}\\ \nonumber \\
\nu_r^{(2)}(p,\lambda=1)= & & { n^2\over \left(16\pi^2\right)^2 } \left [ -8\ln{p^2\over \bar{\mu}^2 } + { 577\over 18}-6\zeta(3)\right]+\nonumber \\ \nonumber \\ & & { n_f\over \left(16\pi^2\right)^2 } \left [ n\left ( 3 \ln{p^2\over \bar{\mu}^2 } - { 401\over 36}\right)+ { 1\over
n } \left(-\ln{p^2\over \bar{\mu}^2 } + { 55\over 12 } - 4\zeta(3)\right)\right ] \label{nu2b}\end{aligned}\ ] ] for the lattice quantities , the gluonic contributions ( @xmath100 ) have been presented in previous works @xcite ( for the wilson action ) @xmath101 the fermionic contributions are associated with the diagrams of fig . 1 and fig .
2 . in the present work
, @xmath102 is perturbatively calculated for the first time using overlap fermions and wilson gluons . for completeness
, we also compute the coefficient @xmath93 and compare it with previous results . the 1-loop diagrams ( fig .
1 ) correspond to @xmath93 , while the 2-loop diagrams ( fig . 2 ) lead to @xmath102 .
dashed lines ending in a cross represent the background field , while those inside loops denote the quantum field .
solid lines correspond to fermion fields and a dot stands for the mass counterterm .
note that , for overlap fermions , the mass counterterm equals zero , by virtue of the exact chiral symmetry of the overlap action ; consequently , diagrams 19 and 20 both vanish .
certain 2-loop diagrams have infrared divergences and become convergent only when grouped together ( 6 + 12 , 7 + 11 , 8 + 18 , 9 + 17 )
. + .8 cm .25 cm
in recent years , overlap fermions are being used ever more extensively in numerical simulations , both in the quenched approximation and beyond .
this fact , along with the desirable properties of the overlap action , was our motivation to calculate the @xmath9-function with this type of fermions .
the important advantage of the overlap action is that it preserves chiral symmetry while avoiding fermion doubling .
it is also @xmath103 improved .
the main drawback of this action is that it is necessarily non - ultralocal ; as a consequence , both numerical simulations and perturbative studies are extremely difficult and demanding ( in terms of human , as well as computer time ) .
the overlap action is given by @xcite @xmath104 where @xmath105 is the overlap - dirac operator @xmath106 , \qquad x = \frac{1}{a^4 } \left(d_w -\rho \right)\end{aligned}\ ] ] and @xmath107 is the wilson - dirac operator @xmath108 , \qquad \nabla_\mu\psi(x ) = { 1\over a } \left [ u(x,\mu ) \psi(x + a\hat{\mu})- \psi(x)\right ] \label{derivative}\ ] ] the overlap parameter @xmath7 is restricted by the condition @xmath109 to guarantee the correct pole structure of @xmath110 .
the coupling constant is included in the link variables , present in the definition of @xmath111 , and one must take the perturbative expansion of @xmath111 in powers of @xmath0 .
this expansion in momentum space takes the form @xmath112 where @xmath113 is the inverse fermion propagator and @xmath114 are the vertices of the wilson fermion action ( @xmath115 ( @xmath116 ) : fermion ( antifermion ) momentum ) . the construction of all overlap vertices relevant to the present computation ( see eqs .
( [ vertices ] ) , ( [ v1 ] ) , ( [ v2 ] ) , ( [ v3 ] ) , ( [ v4 ] ) below ) make use of @xmath113 and @xmath117 ; these quantities can be written in the compact form @xmath118 \nonumber\\ & & x_4(p',p)=\frac{g_0 ^ 4}{4!}\int \frac{d^4k_1d^4k_2d^4k_3d^4k_4}{(2\pi)^{12}}\delta(p'-p-\sum_{i=1}^4 k_i)\sum_{\mu}\prod_{i=1}^4 a_{\mu}(k_i ) \big[-a^2 v_{2,\mu}\big(\frac{p'+p}{2}\big ) \big ] \label{x's}\end{aligned}\ ] ] where @xmath119 @xmath120 represents a gluon field ; later on we will have to generalize eqs .
( [ x s ] ) to the case where both background and quantum gluon field are present , see eq .
( [ x_qa ] ) . at this point
we can proceed with the perturbative expansion of @xmath110 in powers of @xmath0 .
this leads to the propagator of zero mass fermions and to gluon - fermion - antifermion vertices ( with up to 4 gluons for the needs of the present work ) .
the much simpler case of vertices with up to 2 gluons ( and no background ) can be found in ref . @xcite .
after laborious analytical manipulations ( an essential step is the expansion of @xmath121 using complex analysis , which is presented in appendix a ) , the overlap - dirac operator is expanded into terms with up to 4 gluons as @xmath122 @xmath123 is the inverse propagator , @xmath124 and @xmath125 where we have set @xmath126 .
@xmath127,@xmath128 , @xmath129 are given below and the reader can find the expression for @xmath130 in appendix a. @xmath131 } \label{v1}\\ \hskip -.75 cm \phantom{--- } \nonumber \\
\hskip -.75 cm \phantom{--- } \nonumber \\
\hskip -.75 cm v_2^i({k_1}&&,k_2 ) = \int\frac{d^4k_3}{(2\pi)^4}\,\brown{\frac{1}{\omega(k_1)+\omega(k_3)}\frac{1}{\omega(k_1)+\omega(k_2 ) } \frac{1}{\omega(k_2)+\omega(k_3)}}\times \nonumber \\
\hskip -.75 cm & & \sum_{\{j>0,k>0\ } \atop \{j+k = i\ } } \red{\bigg[}-x_j(k_1,k_3)\,\chi_0^{\dagger}(k_3)\,x_k(k_3,k_2)\nonumber \\[-3.0ex ] \hskip -.75 cm & & \qquad\quad\quad - x_j(k_1,k_3)\,x_k^{\dagger}(k_3,k_2)\,\chi_0(k_2)\nonumber \\
\hskip -.75 cm & & \qquad\quad\quad-\chi_0(k_1)\,x_j^{\dagger}(k_1,k_3)\,x_k(k_3,k_2)\nonumber \\
\hskip -.75 cm & & \qquad\quad\quad+\brown{\frac{\omega(k_1)+\omega(k_2)+\omega(k_3)}{\omega(k_1)\omega(k_2)\omega(k_3 ) } } \chi_0(k_1)x_j^{\dagger}(k_1,k_3)\chi_0(k_3)x_k^{\dagger}(k_3,k_2)\chi_0(k_2)\red{\bigg ] } \label{v2}\\ \hskip -.75 cm \phantom{--- } \nonumber \\
\hskip -.75 cm \phantom{--- } \nonumber \\
\hskip -.75 cm v_3^i({k_1}&&,k_2 ) = \int\int{d^4 k_3\over(2\pi)^4}{d^4k_4\over(2\pi)^4 } { 1\over4}\brown{\biggl(\prod_{p \epsilon s_4}{1\over\omega(k_{p_1})+\omega(k_{p_2})}\biggr)}\times\nonumber\\ \hskip -.75 cm & & \sum_{\{j>0,k>0,l>0\ } \atop \{j+k+l = i\ } } \red{\biggl [ } -\frac{1}{6}\brown{\biggl(\sum_{p \epsilon s_4 } \omega(k_{p_1})\omega(k_{p_2})\omega(k_{p_3})\biggr ) } x_j(k_1,k_3)x_k^{\dagger}(k_3,k_4)x_l(k_4,k_2 ) \nonumber\\ \hskip -.75 cm & & + \brown{\bigl(\omega(k_1)+\omega(k_3)+\omega(k_4)+\omega(k_2)\bigr)}\times\nonumber\\ \hskip -.75 cm & & \blue{\bigl[}\chi_0(k_1)x_j^{\dagger}(k_1,k_3)x_k(k_3,k_4)x_l^{\dagger}(k_4,k_2)\chi_0(k_2)+ \chi_0(k_1)x_j^{\dagger}(k_1,k_3)x_k(k_3,k_4)\chi_0^{\dagger}(k_4)x_l(k_4,k_2)\nonumber\\ \hskip -.75 cm & & + \chi_0(k_1)x_j^{\dagger}(k_1,k_3)\chi_0(k_3)x_k^{\dagger}(k_3,k_4)x_l(k_4,k_2)+ x_j(k_1,k_3)\chi_0^{\dagger}(k_3)x_k(k_3,k_4)x_l^{\dagger}(k_4,k_2)\chi_0(k_2)\nonumber\\ \hskip -.75 cm & & + x_j(k_1,k_3)\chi_0^{\dagger}(k_3)x_k(k_3,k_4)\chi_0^{\dagger}(k_4)x_l(k_4,k_2)+ x_j(k_1,k_3)x_k^{\dagger}(k_3,k_4)\chi_0(k_4)x_l^{\dagger}(k_4,k_2)\chi_0(k_2)\blue{\bigr]}\nonumber\\ \hskip -.75 cm & & -\brown{\biggl(\sum_{p \epsilon s_4 } { \omega(k_{p_1})\omega(k_{p_2 } ) \bigl(\omega(k_{p_1})/2+\omega(k_{p_3})/3 \bigr ) \over\omega(k_1)\omega(k_3)\omega(k_4)\omega(k_2)}\biggr)}\times \nonumber\\ \hskip -.75 cm & & { \chi_0(k_1)x_j^{\dagger}(k_1,k_3)\chi_0(k_3)x_k^{\dagger}(k_3,k_4)\chi_0(k_4)x_l^{\dagger}(k_4,k_2)\chi_0(k_2)}\red{\biggr ] } \label{v3}\end{aligned}\ ] ] ( @xmath132 : permutation group of the numbers @xmath133 ) the use of the background field technique implies that instead of the generic gluonic fields ( appearing in @xmath114 s , eq .
( [ x s ] ) ) , one must consider all possible combinations of background ( ) and quantum ( ) fields which originate in the links of eqs .
( [ derivative ] ) , ( [ link ] ) .
hence , @xmath134 as an example , let us write the expression for @xmath135 @xmath136\end{aligned}\ ] ] upon substituting the expression for @xmath114 s in the overlap vertices , the latter become extremely lengthy and complicated .
for instance , the vertex with q - q - a-@xmath137-@xmath138 consists of 9,784 terms , while the vertex with q - q - a - a-@xmath137-@xmath138 has 724,120 terms .
for the algebra involving lattice quantities , we make use of our symbolic manipulation package in mathematica , with the inclusion of the additional overlap vertices .
the first step to evaluate the diagrams is the contraction among vertices , a step performed automatically once the vertices and the ` incidence matrix ' of the diagram are specified .
the outcome of contraction is a preliminary expression for the diagram under study ; there follow simplifications of the color dependence , @xmath139 matrices and tensor structures .
we use symmetries of the theory ( permutation symmetry and lattice rotational invariance ) , or any other additional symmetry that may appear in particular diagrams , to keep the size of the expression down to a minimum .
the external momentum @xmath115 appears in arguments of trigonometric functions and the extraction of @xmath115 dependence is divided into two parts : first we isolate terms that give single and double logarithms ( a few thousand terms , expressible in terms of known , tabulated integrals ) , and then for the convergent terms we employ naive taylor expansion with respect to @xmath115 up to @xmath140 .
this extraction makes explicit the functional dependence of each diagram on @xmath115 ; the coefficients of terms proportional to @xmath141 are integrals over the two internal momentum 4-vectors .
the required numerical integrations are performed by optimized fortran programs which are generated by our mathematica ` integrator ' routine .
each integral is expressed as a sum over the discrete brillouin zone of finite lattices , with varying size @xmath22 , and evaluated for different values of the overlap parameter @xmath7 .
the average length of the expression for each diagram , after simplifications , is about 2 - 3 hundred thousand terms , so that diagrams must be split into parts ( usually of 2000 terms ) to be integrated .
the numerical values of these parts must then be added together to avoid running into systematic errors or spurious divergences .
finally , we extrapolate the results to @xmath142 ; this procedure introduces an inherent systematic error , which we can estimate quite accurately .
infrared divergent diagrams must be summed up before performing the extrapolation .
we denote the contribution of the @xmath143 1-loop feynman diagram to @xmath144 as @xmath145 ; similarly , contributions of 2-loop diagrams to @xmath146 are indicated by @xmath147 .
the quantities @xmath145 , @xmath147 depend on @xmath5 , @xmath4 , @xmath7 and @xmath148 according to the following formulae ( @xmath97 ) @xmath149 \label{nu1lattice}\\ \nonumber \\ & & \widehat{ap}^2 \nu_i^{(2)}(p ) = n_f \biggl[c_{i}^{(0 ) } + a^2 p^2 \left\{c_{i}^{(1)}+ c_{i}^{(2)}\ , { \ln a^2 p^2 \over ( 4 \pi)^2 } + c_{i}^{(3 ) } \left({\ln a^2 p^2 \over ( 4 \pi)^2}\right)^2 + c_{i}^{(4)}{\displaystyle\sum_\mu p_\mu^4 \over ( p^2)^2 } \right \ } + { \cal o}((ap)^4)\biggr ] \label{nu2lattice}\end{aligned}\ ] ] where @xmath150 .
the index @xmath151 runs over diagrams , and the coefficients @xmath152 , @xmath153 depend on the overlap parameter @xmath7 .
moreover , @xmath154 $ ] .
comparison with continuum results and usage of ward identities requires @xmath155 we have checked that all the above conditions are verified by our results . inserting these conditions in eqs .
( [ nu1lattice ] ) , ( [ nu2lattice ] ) , the expressions for the fermionic contribution to @xmath144 , @xmath146 ( after addition of all diagrams ) take the form @xmath156
\label{nu1lattice_final}\\ \nonumber \\ & & \nu^{(2)}(p ) = \nu^{(2)}(p)\big|_{n_f=0 } + n_f \biggl[\sum_i \big({c_{i}^{(1,1)}\over n } + n c_{i}^{(1,1 ) }
\big ) + { 1\over 16 \pi^2 } ( 3 n - { 1\over n})\ , { \ln a^2 p^2 \over ( 4 \pi)^2 } + { \cal o}((ap)^2)\biggr ] \label{nu2lattice_final}\end{aligned}\ ] ] in table i we tabulate the total 1-loop contribution @xmath157 for 21 values of the overlap parameter ( @xmath8 ) .
each diagram was integrated for lattice size @xmath158 , @xmath159 and then extrapolated to @xmath160 . in all tables and figures ,
the errors accompanying our results are entirely due to this extrapolation .
the coefficients @xmath161 do not depend on the number of colors @xmath5 , nor on the choice of regularization for the pure gluonic part of the action ( symanzik , iwasaki , etc . ) .
the numbers in table i are in agreement with corresponding numbers from ref .
the 2-loop calculation of @xmath146 was accomplished for the same 21 values of @xmath7 and for @xmath162 .
table ii presents the coefficients @xmath163 , @xmath164 , versus @xmath7 . due to the extremely large size of the vertices involved
, it is almost impossible to extend the results to larger @xmath22 .
typically , the integration of 2000 terms is completed in @xmath17 7 days on 1 cpu ; the present calculation comprises approximately 3500@xmath1652000 terms .
thus , if only a single cpu were available , our work would have required @xmath1750 years . in certain cases with large systematic errors we extended the integration up to @xmath166 . in general
, the overlap action leads to coefficients which are very small for most values of @xmath7 . as a consequence , systematic errors , which are by and large rather small ,
tend to be significant fractions of the signal for @xmath167 . from eqs .
( [ zg ] ) , ( [ zg_expanded])-([omega1_1 ] ) , ( [ nu2_1 ] ) we find the following expression for @xmath168 in terms of @xmath169 @xmath170 , @xmath171 , @xmath172 @xmath173 \label{l1}\end{aligned}\ ] ] we can write the final form of the 3-loop coefficient @xmath47 for the @xmath9-function ( eq .
( [ b2lrel ] ) ) , including gluonic as well as fermionic contributions , using eqs .
( [ univ_coeff ] ) , ( [ b2ms ] ) , ( [ l0 ] ) and ( [ l1 ] ) @xmath174 \label{b2_final}\end{aligned}\ ] ] a large variety of possible numerical checks has been performed , as mentioned above : * a. * the total contribution to the gluon mass adds to zero , as expected .
* b. * the coefficients of the non - lorentz invariant terms cancel . * c. * the terms with double logarithms correspond to the continuum counterparts .
this has been checked diagram by diagram . *
d. * terms with single logarithms add up to their expected value , which is independent of @xmath7 ( although the expressions per diagram are @xmath7-dependent ) . in fig .
3 we plot the 1-loop coefficient @xmath175 with respect to @xmath7 .
note that the errors are too small to be visible at this scale .
the 2-loop coefficients @xmath176 and @xmath177 are plotted in figs .
4 - 5 , respectively , for different values of the overlap parameter .
the extrapolation errors are visible for @xmath178 and @xmath179 . substituting @xmath175 , @xmath176 and @xmath177 into eq .
( [ b2_final ] ) , we find the numerical results for the 3-loop contribution , @xmath47 , of the @xmath9-function .
these are plotted in fig .
6 , choosing @xmath180 and @xmath181 .
-5 mm fig .
3 : plot of the total 1-loop coefficient @xmath182 versus the overlap parameter @xmath7 .
.5 cm fig .
4 : plot of the total 2-loop coefficient @xmath183 versus @xmath7 .
.5 cm fig . 5 : plot of the total 2-loop coefficient @xmath184 versus @xmath7 .
.5 cm .5 cm fig . 6 : the 3-loop coefficient @xmath47 ( eq .
( [ b2_final ] ) ) , plotted against @xmath7 , for @xmath180 and @xmath185 ( horizontal red line ) , @xmath186 ( green line ) and @xmath187 ( blue line ) .
we have calculated the 2-loop coefficient of the coupling renormalization function @xmath6 , for the yang - mills theory with gauge group @xmath188 and @xmath4 species of overlap fermions .
we used the background field method to simplify the computation ; in this method there is no need of evaluating any 3-point functions .
this is the first 2-loop calculation using overlap fermions with external momenta , and it proved to be extremely demanding in human and cpu time ; this is due to the fact that we had to manipulate very large expressions ( millions of terms ) in intermediate stages .
we used our numerical results of @xmath6 to determine the 3-loop coefficient @xmath47 of the bare lattice @xmath9-function ; the latter dictates the asymptotic dependence between the bare coupling constant @xmath0 and the lattice spacing @xmath64 , required to maintain the renormalized coupling at a given scale fixed .
knowledge of @xmath47 provides the correction term to the standard asymptotic scaling relation between @xmath64 and @xmath0 , via eq .
( [ asympre ] ) .
the dependence of @xmath6 and @xmath47 on @xmath5 and @xmath4 is shown explicitly in our expressions .
on the other hand , dependence on the overlap parameter @xmath7 can not certainly be given in closed form ; instead , we present our results for a large set of values of @xmath7 in its allowed range . the 3-loop correction is seen to be rather small : this indicates that the perturbative series is very well behaved in this case , despite the fact that it is only asymptotic in nature .
furthermore , around the values of @xmath7 which are most often used in simulations ( @xmath189 ) , fermions bring about only slight corrections to the 3-loop @xmath9-function , even compared to pure gluonic contributions , as can be seen from fig .
the only source of numerical error in our results has its origin in an extrapolation to infinite lattice size .
compatibly with the severe cpu constraints , numerical 2-loop integration had to be performed on lattices typically as large as @xmath190 , or even up to @xmath191 in cases where an improved extrapolation was called for .
an intermediate range for @xmath7 ( @xmath192 ) showed the most stable extrapolation error , and this may be a sign of their suitability for numerical simulations .
as a by product of the present work , we have produced the lengthy expressions corresponding to _ all _ overlap vertices which can arise in a 2-loop computation , and presented them in a rather compact form .
further computations of similar complexity , for example the 2-loop renormalization of operators in the overlap action ( such as fermion currents ) , only require the vertices which we have presented here .
* acknowledgements : * this work is supported in part by the research promotion foundation of cyprus ( proposal nr : @xmath193/0505/45 ) .
[ tab1 ] .numerical results for the 2-loop coefficients @xmath194 and @xmath195 [ cols="^ , < " , ] .25 cm
here we will explain the basic idea of expanding the expressions @xmath121 , appearing in the overlap - dirac operator , in powers of @xmath0 , using a procedure introduced by y. kikukawa and a. yamada @xcite .
we also provide the expression for @xmath196 further below . in an integral representation
, the combination @xmath197 can be written as @xmath198 in fact , eq .
( [ xxdag ] ) is valid for any operator @xmath111 provided that @xmath199 has no vanishing eigenvalues .
we begin the desired expansion of the overlap - dirac operator in powers of @xmath0 , by setting @xmath200 the first term of eq .
( [ xxdag_expansion ] ) corresponds to the inverse fermionic propagator , while @xmath201 leads to the vertices ; for our 2-loop calculation we need to write @xmath201 up to @xmath202 @xmath203 using the above equations , we write the denominator on the r.h.s . of eq .
( [ xxdag ] ) as @xmath204 from this point forward it is easier to work in momentum space since , taking the fourier transform , the denominator becomes diagonal @xmath205 = \frac{1}{t^2 + \omega^2(p)}\ ] ] ( @xmath206 defined in eq .
( [ d0_omega ] ) ) . combining eqs .
( [ xxdag ] ) and ( [ txxdag ] ) we derive the taylor expansion of @xmath197 in momentum space @xmath207 the integral over @xmath208 can now be performed by closing the contour around the upper complex @xmath208-plane .
considering , as an example the third term on the r.h.s .
( [ sqrtxxdag_momentum ] ) , the result is @xmath209\left[\omega(k){+}\omega(p)\right]\left[\omega(p){+}\omega(p')\right]}\ ] ] similarly we integrate all terms of eq .
( [ sqrtxxdag_momentum ] ) over @xmath208 and this leads to eqs .
( [ d0_omega])-([v3 ] ) and to the expression for @xmath196 which is presented below ( @xmath210 : permutation group of the numbers 1 through 5 ) | we calculate the relation between the bare coupling constant @xmath0 and the @xmath1-renormalized coupling @xmath2 , @xmath3 , to 2 loops in perturbation theory .
we employ the standard wilson action for gluons and the overlap action for fermions . for convenience ,
we have worked with the background field technique , which only requires evaluation of 2-point green s function for the problem at hand .
our results depend explicitly on the number of fermion flavors ( @xmath4 ) and colors ( @xmath5 ) .
since the dependence of @xmath6 on the overlap parameter @xmath7 can not be extracted analytically , we tabulate our results for different values in the allowed range of @xmath7 ( @xmath8 ) , focusing on values which are being used most frequently in simulations .
knowledge of @xmath6 allows us to derive the 3-loop coefficient of the bare @xmath9-function ( @xmath10 ) which , unlike the 1- and 2-loop coefficients , is regularization - dependent .
the nontrivial dependence of @xmath6 and of @xmath10 on @xmath7 is plotted for various choices of @xmath5 , @xmath4 .
beta function , overlap fermions , running coupling constant , lattice perturbation theory . 11.15.ha , 12.38.gc , 12.38.bx , 11.10.gh . |
the hexatriynyl radical c@xmath3h , a member of the linear c@xmath5h series , belongs to the spectroscopically best studied carbon chain radicals .
the x@xmath1 ground state was studied extensively in fourier - transform microwave ( mw ) studies , yielding accurate constants @xcite .
the laboratory data followed astronomical detections of this molecule in the cold dense cloud tmc-1 @xcite .
later this molecule was detected along other lines of sight as well , e.g. towards the carbon rich star irc+10216 @xcite .
the b@xmath0x@xmath1 electronic spectrum of c@xmath3h was reported in several studies .
the first data were recorded in ne matrix isolation studies after mass selective deposition ( see e.g. * ? ? ? * ) .
this provided approximate values for the gas phase absorptions .
the first gas phase spectrum was obtained under somewhat poor experimental conditions , using a hollow cathode discharge cell @xcite .
the resulting spectrum only showed unresolved features , including many bands starting from vibrationally excited states .
subsequently , high precision and rotationally resolved spectra were recorded using a new planar plasma source design @xcite that allowed the detection of rotationally cold c@xmath3h and c@xmath3d transients under nearly doppler - free conditions @xcite . combined with sensitive detection techniques , such as cavity ring - down spectroscopy
, it was possible to realize excellent s / n ratios , not only for the origin band but also for transitions involving vibrationally excited levels in the upper electronic state . in combination with the available mw constants it was possible to derive precise upper state constants , for both spin - orbit components .
recently , a substantially extended study of these electronic bands , as well as new transitions involving vibronic hot bands was presented , with a focus on renner - teller effects in c@xmath3h @xcite . in the present work
we extend previous spectroscopic studies to the fully @xmath2c - substituted isotopologue @xmath2c@xmath3h .
the rotationally resolved electronic spectrum yields accurate spectroscopic constants , also including the ground state constants that have not been determined before .
we follow an established method that has been used successfully to record electronic spectra of many different carbon chain radicals before .
details of the experimental setup and experimental procedures are available from @xcite .
supersonically jet - cooled @xmath2c@xmath3h radicals are produced among other species in a pulsed ( @xmath610 hz ) planar or pinhole plasma expansion generated by discharging a high pressure ( 10 bar ) gas pulse comprising of 0.18% @xmath2c - enriched ( 99% purity ) c@xmath4h@xmath4 in a helium : argon @xmath685:15 mixture . both slit and pinhole discharge nozzles , which have been used to study @xmath7c@xmath3h in ref .
@xcite , are employed in the present experiment . the 3 cm @xmath8 300 @xmath9
m slit discharge nozzle , with typical discharge i / v values of @xmath6100 ma/@xmath10750 v per jaw , is used to generate a planar plasma expansion with a nearly doppler - free environment and a relatively long effective absorption path length .
the pinhole nozzle system , with typical i / v @xmath680 ma/@xmath101000 v values , generates a plasma expansion suffering from a somewhat larger doppler width , but allows measuring spectra with different rotational temperatures in a more convenient way .
final rotational temperatures typically amount to 15 - 30 k. the @xmath2c@xmath3h spectra are recorded in direct absorption using cavity ring - down laser spectroscopy .
tunable light around 525 nm the region where regular c@xmath3h has been found is generated using a nd : yag pumped pulsed dye laser ( @xmath66 ns pulse duration ) .
the bandwidth of the dye laser is @xmath60.035 @xmath11 , well below the expected line splitting between two adjacent rotational transitions .
the light pulse is spatially filtered and enters a 58 cm long optical cavity comprising of two high reflection mirrors .
the optical axis is aligned a few mm downstream , parallel to the slit or below the pinhole of the nozzle .
light leaking out of the cavity is recorded by a photo - multiplier tube and a spectrum is obtained by recording the ring - down time as function of the laser wavelength
. typical ring - down times amount to 40 - 100 @xmath9s .
a precise pulse generator coordinates that the gas and the discharge pulse , and as well as the ring - down event , coincide . simultaneously with the spectral recordings of @xmath2c@xmath3h , transmission fringes of an etalon ( with a free spectral range of @xmath620.1 ghz ) are recorded , providing relative frequency markers to correct for a possible non - linear wavelength scanning of the dye laser .
an iodine spectrum allows an absolute wavelength calibration with an accuracy better than 0.02 @xmath11 .
in figure 1 the b@xmath0x@xmath1 electronic origin band ( @xmath12 ) transition of regular c@xmath3h ( upper panel ) and of @xmath2c@xmath3h ( lower panel ) is shown .
the newly recorded @xmath7c@xmath3h spectrum compares well with data presented in @xcite . from a comparison of both panels
it can be concluded that the @xmath2c - sample is highly pure , as nearly no features due to @xmath7c contamination can be clearly seen .
some extra lines in the lower panel recorded through the slit expansion
are due to overlapping hot band transitions of other small hydrocarbons in the plasma .
previous work has shown that these lines become much weaker using the pinhole discharge nozzle , as visible from the corresponding spectrum .
it should be noted that in contrast to the pinhole nozzle experiments in @xcite , the final doppler width of individual transitions is smaller , mainly because of the reduced expansion velocity of the jet that is run with more argon in the gas mixture .
the overall spectral pattern of @xmath2c@xmath3h is very similar to that of @xmath7c@xmath3h , with a blue shift of @xmath62 @xmath11 .
both spectra consist of two components , each with ( fully resolved ) p , q and ( partially resolved ) r branches , reflecting transitions between the two spin - orbit components : b@xmath13x@xmath14 and b@xmath15x@xmath16 .
the two band systems are split by @xmath17a = @xmath18a@xmath19a@xmath20 , the difference between the spin - orbit splitting in ground and excited state .
an assignment of the individual spin - orbit components is straightforward ; with a low final temperature in the jet expansion , the more intense component corresponds to a transition from the lower spin - orbit state , i.e. x@xmath14 for c@xmath3h .
the observation of the barely visible q - branch in the b@xmath15x@xmath16 system provides further evidence for this assignment , as the intensity of the q - branch scales quadratically with the value of the spin - orbit angular momenta ( @xmath21 ) .
the detailed rotational assignments for @xmath2c@xmath3h are shown in figure 1 .
the determined transition energies of the 116 assigned lines listed in table 1 are used to fit five parameters . for regular c@xmath3h
a full analysis has been presented before @xcite , guided by accurate ground state constants available from mw work @xcite .
the focus here is on @xmath2c@xmath3h for which ground state constants are not available ; it is important , therefore , to realize that ground and excited state parameters may correlate in a fit .
[ cols= " > , > , > , > , > , > , > , > , > , > , > , > , > , > " , ] as no mw data for the ground state exists , it is wise to check for possible correlations between the resulting ground and upper state parameters using a global fit .
indeed , the fitted values for a and b give high correlation coefficients ( 1.000 and 0.997 , respectively ) .
for this , combination differences for both ground and excited state have been fitted separately .
the resulting molecular parameters are summarized as fit - ii in table 2 .
as can be seen , the derived values for ( b@xmath22 , a@xmath22 ) and ( b@xmath23 , a@xmath23 ) are in good agreement with those derived from fit - i .
a simulated spectrum using these constants is incorporated in the lower panel of figure 1 .
the overall pattern reproduces well for a rotational temperature of t~rot~ = 20.3(1 ) k. this not only applies to the boltzmann contour , but also to the relative intensity ratio of the two spin - orbit components , indicating that the spin - orbit relaxation is as effective as the rotational cooling .
the fitted band origin of the b@xmath0x@xmath1 band of @xmath2c@xmath3h is roughly 2.35 @xmath11 blue shifted with respect to the main isotopologue .
the rotational constant b@xmath22(@xmath2c@xmath3h ) is found to be 7.4% smaller than that of b@xmath22(@xmath7c@xmath3h ) , fully consistent with predictions from the previously determined c@xmath3h structure in ref
. @xcite . upon electronic excitation
, the value of b@xmath23 decreases , indicating that the molecule stretches , but as the inter - atomic forces are expected to barely change , the b@xmath23/b@xmath22 ratios for @xmath24c@xmath3h are nearly the same .
it should be noted that the effective spin - orbit splitting constants ( @xmath18a@xmath20 and @xmath18a@xmath25 ) for @xmath2c@xmath3h are smaller than those for @xmath7c@xmath3h .
this may be due to a different renner - teller effect ( i.e. , the electronic orbital - vibration interaction ) in @xmath2c@xmath3h .
indeed , spin - orbit coupling can be significantly affected by renner - teller effects and as discussed in ref .
@xcite , for @xmath7c@xmath3h a very strong renner - teller effect exists .
this work has been supported by the netherlands foundation for fundamental research on matter ( fom ) and the netherlands organisation for scientific research ( nwo ) through a vici grant and within the context of the dutch astrochemistry network ( dan ) .
x.b . and w.u .
acknowledge support from the templeton foundation . | the rotationally resolved spectrum of the b@xmath0x@xmath1 electronic origin band transition of @xmath2c@xmath3h is presented .
the spectrum is recorded using cavity ring - down spectroscopy in combination with supersonic plasma jets by discharging a @xmath2c@xmath4h@xmath4/he / ar gas mixture . a detailed analysis of more than a hundred fully - resolved transitions allows an accurate determination of the spectroscopic parameters for both the ground and electronically excited state of @xmath2c@xmath3h .
hexatriynyl radical , cavity ring - down spectroscopy , isotopic substitution , supersonic plasma discharge |
intermediate - mass black holes ( imbhs ) are thought to occupy the mass gap between the well - established stellar - mass bhs with @xmath7 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and the well - studied supermassive bhs with @xmath8 ( e.g. , * ? ? ?
finding imbhs in the local universe can provide important insight into formation channels for seed bhs in the early universe ( for reviews , see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . here , we focus on theoretical predictions that globular clusters ( gcs ) could host imbhs formed via dynamical processes stemming from the clusters closely packed stars ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in addition , if imbhs exist in gcs , inspirals involving them and stellar - mass bhs could be years - long sources of gravitational waves for a lisa - like mission @xcite .
such imbhs could also be used to test scaling relations between central bhs and their stellar system hosts , thus informing the debate about whether or not these entities co - evolve ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a key science driver for future near - infrared , ground - based telescopes is to measure , at a distance of 10 mpc , a bh mass as low as @xmath9 by spatially resolving its sphere of influence in its host stellar system @xcite .
it is expected that this will yield a robust inventory of imbhs in gcs in the local universe , but these next - generation facilities are many years off . moreover , although gcs in the local group have been targeted in sphere - of - influence studies , all these studies are contentious ( * ? ? ? * and references therein ) . given these controversies and long waits for future facilities ,
we have begun an independent investigation , namely searching for radio signatures of accretion onto putative imbhs in extragalactic gcs @xcite . by analogy with stellar - mass bhs ( reviewed by * ? ? ?
* ) , one expects that an imbh will spend more time in the hard x - ray state - including quiesence - associated with a low accretion rate onto the bh , than in the soft x - ray state associated with a high accretion rate .
in the typical case of only a few radio observations , it is likely that they will sample the steady radio emission characteristic of the hard x - ray state , as opposed to the flaring radio emssion associated with a transition from the hard x - ray state to the soft x - ray state .
these concepts , first laid out by @xcite , lead to the following three approaches : ( 1 ) : : detect radio emission like that from hlx-1 in its hard x - ray state , where hlx-1 is a strong imbh candidate in an extragalactic star cluster with a stellar mass of @xmath10 @xcite .
( 2 ) : : use the empirical fundamental - plane regression for the hard x - ray state , plus observations of x - ray and radio luminosities , to estimate an imbh mass @xcite . ( 3 ) : : use a conservative , semi - empirical model to predict the mass of an imbh that , if experiencing bondi accretion in the hard x - ray state , would be consistent with the observed radio luminosity ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
our radio imaging of ngc1023 , an early - type galaxy about 11 mpc away , detected none of its 337 candidate star clusters @xcite .
the approaches above then led to the following inferences : ( 1 ) no hlx-1 analogs suggests that phenomenon is very rare or is fuelled from gas related to its cluster s relatively young stars .
( 2 ) to be able to reach the regime of imbh masses , deeper x - ray and radio surveys that detect candidate clusters at lower luminosities are needed .
importantly , the mass term in the fundamental plane implies that , in the hard - x - ray state , the radio emission from imbhs should be much brighter than that from stellar - mass bhs .
this makes radio detections effective at filtering out contamination from the x - ray - emitting stellar - mass bhs often found in star clusters .
( 3 ) the radio - luminosity upper limit for a stack of the 20 most massive clusters corresponds to a mean 3@xmath2 imbh mass of @xmath11 and a bh mass fraction @xmath12 .
the inferences from our ngc1023 study can be improved upon by obtaining longer exposures on star clusters in closer galaxies .
we thus turn to , a spiral galaxy at a distance of [email protected] mpc ( 1 = 17.6 pc ; * ? ? ?
* ) that is estimated to have 210@xmath1330 gcs in total @xcite .
several studies have used the _ hubble space telescope ( hst ) _ to localize candidate gcs in m81 @xcite . here , we focus on the @xcite study because it leveraged advanced camera for surveys ( acs ) mosaics with spectroscopy to identify probable gcs in m81 . @xcite performed aperture photometry and profile fitting on 419 candidate gcs , and highlight 214 as being probable gcs because they were spectroscopically confirmed ( 85 ) or were good candidates ( 129 ) that shared the color and size ranges of the spectroscopically confirmed gcs . the v - band ( f606w ) magnitudes from @xcite and that band s mass - to - light ratio for gcs from @xcite imply that the stellar masses of the 214 probable gcs range from @xmath14 to @xmath15 . in this paper
, we use the nrao karl g. jansky very large array ( vla ; * ? ? ?
* ) to search for radio emission from these probable gcs in m81 .
we describe our new vla results in [ imaging ] .
no individual probable gc is detected .
the implications of these nondetections are then explored regarding hlx-1 analogs ( [ hlx ] ) , x - ray detected clusters ( [ x - ray ] ) , and semi - empirical model predictions ( [ model ] ) .
we close in [ sumcon ] with a summary and conclusions .
we observed m81 under proposal code 13b-138 ( pi m. middleton ) using the vla in its b configuration at a central frequency of 5.5 ghz , corresponding to a wavelength of 5.5 cm .
owing to the large angular size of the galaxy , we used four pointing centers separated by 45 to provide relatively uniform sensitivity to the central regions of the galaxy .
the pointing centers locations and observation dates appear in table 1 .
each vla pointing was observed for one hour , giving 41 m of time on target , using a correlator integration time of 3 s. the target elevation lay between 23@xmath16 and 32@xmath16 in all cases .
the observing bandwidth was split into two 1024-mhz basebands centred at 5.0 and 6.0 ghz , each comprised of eight spectral windows of width 128 mhz , each of which was split into sixty - four 2-mhz channels .
we used 3c147 to set the amplitude scale to an estimated accuracy of about 3% , and j1048 + 7143 as the secondary calibrator to derive the atmospheric and instrumental complex gains on a per - antenna basis .
the position assumed for j1048 + 7143 was @xmath17 and @xmath18 with one - dimensional errors at 1@xmath2 of 2 mas .
given the observing strategies , the one - dimensional astrometric error at 1@xmath2 is estimated to be 01 .
we processed the data using standard procedures within the common astronomy software application ( casa ; * ? ? ?
* ) release 4.1.0 . following external gain calibration , the stokes @xmath19 data for each pointing
were imaged with the casa task clean using a robustness parameter of 0.5 to obtain the best compromise among sensitivity , spatial resolution and sidelobe suppression ; an nterms parameter of 2 to accommodate the large fractional bandwidth ; and the gridmode parameter set to `` widefield '' and the wprojplanes parameter set to 128 to correct for the effects of non - coplanar baselines .
the images were dominated by the emission from the low - luminosity active galactic nucleus ( llagn ) in m81 , and required self - calibration , initially in phase , and then in amplitude and phase , down to a timescale as short as the correlator integration time of 3 s. the final , self - calibrated images of all four quadrants were restored with the same circular gaussian restoring beam of fwhm of 15 , and then mosaicked together , cutting off the response of the primary beam at the 10% level . in the mosaic , the llagn is point - like and has a flux density of 81@xmath133 mjy .
figure 1 shows the geometry of the vla mosaic overlaid on a @xmath20 image of m81 @xcite retrieved from ned .
for each of the 419 candidate gcs , task subim in the 2016 december 31 release of nrao s astronomical image processing system ( aips ; * ? ? ?
* ) was used to form a cutout spanning 30@xmath21 and centered on the optical position @xcite .
astrometric comparions between the acs images and ground - based surveys implied conservative one - dimensional errors at 1@xmath2 of 02 @xcite .
thirteen candidate gcs lay beyond the boundary of the vla mosaic .
this left 406 candidates for analysis , split as 206 probable gcs and 200 improbable gcs .
the term `` improbable '' refers to candidate gcs that @xcite judged to be background galaxies or non - gc objects like young clusters and hii regions , or confirmed as galaxies or non - gc objects via spectroscopy .
figure 2 shows the vla cutouts for the 206 probable gcs .
the 1@xmath2 rms noise level among the cutouts varies from 4.3 to 51 @xmath22jy beam@xmath23 , a range expected given the primary beam cutoff invoked .
an individual one - tail detection threshold of 3@xmath2 is adopted to minimize the risk of a false - positive detection of one or more targets when examining the ensembles of 206 probable gcs and 200 improbable gcs @xcite .
only one of the 200 improbable gcs , i d 146 from @xcite , was detected with enough significance to serve as a crosscheck of the radio and optical astrometry .
the vla cutout of i d 146 appears in figure 3 .
the radio emission is somewhat resolved but dominantly compact .
the radio peak has a total offset of @xmath24 when compared to the optical position reported by @xcite . assuming that the peaks should physcially coincide , such an offset is consistent with the error estimates cited above for the radio and optical data .
moreover , finding resolved radio emission from i d 146 is consistent with its classification as an improbable gc : if it was a probable gc hosting an imbh in the low - hard state , any radio emission should be spatially unresolved .
from figure 2 , none of the 206 probable gcs in m81 is detected above its local 3@xmath2 level at 5.5 ghz .
the associated radio luminosities are @xmath25 erg s@xmath23 , using a definition that implicitly assumes a flat radio continuum spectrum up to 5.5 ghz .
figure 4 conveys the radio luminosities and stellar masses of each of the 206 probable gcs . at these radio luminosities ,
accreting stellar - mass compact objects in the hard x - ray state would have radio luminosities too faint to be detected @xcite .
what about contamination from flaring radio emssion associated with a transition from the hard x - ray state to the soft x - ray state ?
indeed , such flaring is thought to be the explanation for the stellar - mass bh in m31 that achieved a peak 5-ghz luminosity of @xmath26 erg s@xmath23 and then decayed on a timescale of days @xcite .
however , the lack of radio detections in figure 4 implies that such flaring emission from stellar - mass bhs can not be a major contaminant for these single - epoch observations of the probable gcs in m81 .
( stronger inferences about flares will be reported elsewhere in conjuction with radio monitoring observations . )
this paves the way for us to interpret figure 4 within the context of the three approaches mentioned in
[ motivation ] , namely analogs of hlx-1 ( [ hlx ] ) , x - ray detected clusters ( [ x - ray ] ) , and predictions of a semi - empirical model ( [ model ] ) . in the analysis to follow , it is important to keep in mind that only 40% of the probable gcs have been spectroscopically confirmed @xcite .
hlx-1 is a strong imbh candidate , of mass @xmath27 , in an extragalactic star cluster with an observed stellar mass of @xmath28 @xcite .
the host cluster s observed mass resembles those of the probable gcs in m81 , but the age of the host cluster s light - dominating stars , about 20 myr , is significantly younger than gcs , which are older than 10 gyr @xcite .
still , the host cluster might be more massive and older than deduced so far , if it is actually the remnant of a stripped dwarf galaxy @xcite .
as the latter authors note , such extra mass could also help retain the gas from which the younger stars formed .
near 7 ghz hlx-1 can achieve a flaring luminosity of @xmath29 erg s@xmath23 during transits from its hard to soft x - ray states @xcite , and a steady luminosity of @xmath30 erg s@xmath23 while in its hard x - ray state @xcite . if that steady emission is doppler boosted by a factor of about five to ten , as @xcite argue , its side - on luminosity is about @xmath31 erg s@xmath23 .
given these various radio luminosities for hlx-1 , figure 4 makes it clear that no hlx-1 analog resides in any of the 206 probable gcs in m81 .
@xcite found the same result for 337 candidate gcs in ngc1023 and concluded that ( i ) hlx-1 is accreting gas related to the formation and/or presence of the 20-myr - old stars in its host cluster or ( ii ) the hlx-1 phenomenon is just so rare that no radio analog is expected in ngc1023 .
the same two conclusions apply in the case of m81 .
@xcite used extensive _
chandra _ data to localize several hundred x - ray sources in m81 , and a. zezas ( 2015 , private communication ) is investigating their matches with probable gcs . in the interim , we used version 1.1 of the chandra source catalog @xcite to identify x - ray sources within the optical positional uncertainty of m81 s probable gcs .
we also required that an x - ray detection had enough significance to establish a luminosity in the 0.5 - 10 kev band .
this led to detections of @xcite ids 273 and 284 as sources cxoj095547.0 + 690551 and cxoj095549.7 + 690531 with luminosities of @xmath32 erg s@xmath23 and @xmath33 erg s@xmath23 , respectively .
these detections are consistent with prior _ chandra _ results @xcite .
importantly , @xcite noted that the two x - ray sources , detected in 16 or 17 observations , were always in a hard x - ray state , a requisite for applying the empirical fundamental - plane relation among the x - ray luminosity , @xmath34 , the radio luminosity , @xmath35 , and the bh mass , @xmath36 .
as we wish to estimate masses , we employ the @xcite regression of the contracted sample of @xcite .
inserting the above values for @xmath34 and figure 4 s upper limits to @xmath35 into that regression , we estimate bh masses at 95% confidence of @xmath37 for i d 273 and @xmath6 for i d 284 .
the clusters stellar masses are known from figure 4 , and lead to estimates for the bh mass fractions of @xmath38 for i d 273 and @xmath39 for i d 284 .
we comment further on these estimates in [ model ] .
as cautioned in [ motivation ] , we can not rule out contamination from x - ray - emitting stellar - mass bhs . indeed , that is the more likely explanation because few imbhs are known , whereas many x - ray binaries are associated with extragalactic gcs .
it is also noteworthy that no probable gc in m81 has an x - ray luminosity as high as that of hlx-1 in its hard state , @xmath40 erg s@xmath23 @xcite .
this absence of an x - ray analog of hlx-1 is consistent with the absence of a radio analog of hlx-1 ( [ hlx ] ) .
however , since our limits on the radio luminosity are over an order of magnitude lower than even a de - boosted version of hlx-1 , we are also able to rule out the presence of less extreme sources , even in the case where such a system was heavily absorbed and viewed edge - on , such that the x - ray emission would not be particularly remarkable .
we adopt the semi - empirical model of @xcite to predict the mass of a putative imbh that , if experiencing hard - x - ray - state accretion in a gc , is consistent with the upper limit on the radio luminosity .
following @xcite we conservatively assume that the imbh in a gc accretes at a fraction @xmath41 of the bondi rate from a medium with a gas density @xmath42 @xmath43 .
such values yield a prediction for the hard - state x - ray luminosity @xmath34 .
then , the empirical fundamental - plane regression for the @xcite contracted sample is employed to predict the associated radio luminosity @xmath35 . in this way , a detection of , or upper limit to , a radio luminosity maps to a detection of , or upper limit to , an imbh mass .
figure 5 shows the results of applying this conservative , semi - emipirical model to the 206 probable gcs in m81 .
the best radio luminosity constraint for an individual gc implies a 3@xmath2 imbh mass of @xmath44 . to reach a lower mass regime ,
the aips task stack was used to form a weighted - mean image stack ( e.g. , * ? ? ?
* ) of the cutouts of all 206 probable gcs .
that stack , presented in figure 6 , measures the the mean contribution to the total radio emission from the probable gcs .
for the all - cluster stack , the 3@xmath2 radio - luminosity upper limit corresponds to an imbh mass of @xmath45 .
however , the gcs stellar masses , @xmath46 , range from about 10 thousand to 8 million solar masses , suggesting that it is physically more relevant to focus only on the more massive gcs . like @xcite ,
we thus focus on the gcs whose putative imbhs have masses that are less than half the combined mass of their stars . the 49 gcs meeting that criterion
are listed in table 2 and plotted with dark symbols in figure 5 .
the column contents of table 2 are : column 1 , the i d from @xcite ; column 2 , the stellar mass based on the v - band ( f606w ) magnitudes from @xcite and that band s mass - to - light ratio for gcs from @xcite ; column 3 , the imbh mass based on the semi - empirical model ; and column 4 , the ratio of the imbh mass in column 3 to the stellar mass in column 2 .
each of the 49 massive gcs in table 2 has a stellar mass @xmath47 .
fully 13 of these massive gcs have upper limits to their bh mass fractions , @xmath48 , in the range @xmath49 .
such upper limits are thus below the remarkable value of 0.15 reported for m60-ucd1 @xcite .
the attributes of i d 273 in table 2 deserve special mention : applying the semi - empirical model leads to estimates of a bh mass of @xmath50 and a bh mass fraction of @xmath51 . for comparison ,
applying the empirical fundamental plane , as in [ x - ray ] , led to estimates of @xmath5 and @xmath38 .
we also used the aips task stack to form a weighted - mean image stack of the cutouts of these 49 massive gcs .
figure 7 shows that stack , a measure of the the mean contribution to the total radio emission from the 49 massive gcs .
the stack s 3@xmath2 radio - luminosity upper limit corresponds to an imbh mass of @xmath52 for m81 , a stronger constraint than the upper limit of @xmath53 for ngc1023 @xcite .
the mean stellar mass of the massive gcs in m81 is @xmath54 .
taking the ratio of these values leads to a bh mass fraction @xmath55 for m81 , improving over the equivalent fraction of less than 0.16 for ngc1023 @xcite .
the bh mass fraction of less than 0.08 for the massive gcs in m81 is well below the bh mass fraction of 0.15 for m60-ucd1 @xcite .
still , it should be kept in mind that the interpretation of the bh mass fraction for m81 is affected by unknowns like the distribution function of imbh masses and the fraction of massive gcs occupied by an imbh . within the context of this semi - empirical model , the vla constraints on individual gcs and stacks of gcs are beginning to probe the domain of imbhs in extragalactic gcs .
these constraints for m81 can be improved with longer vla exposures or similar exposures with the next - generation vla ( ngvla ; * ? ? ? * ) . for perspective , assuming a sufficient dynamic range given m81 s llagn , a one - hour exposure with the ngvla at a frequency of 10 ghz could reach an rms noise of 0.45 @xmath22jy beam@xmath23 for an individual gc .
such an ngvla value would be very close to the rms noise achieved in figure 6 only after stacking the vla images of all 206 probable gcs in m81 .
we used a four - pointing vla mosaic at 5.5 ghz to search for the radiative signatures of imbh accretion from 206 probable gcs in m81 , a spiral galaxy at a distance of 3.63 mpc .
none of the individual gcs were detected .
similarly , only upper limits were obtained from weighted - mean image stacks of all 206 gcs and of the 49 massive gcs with @xmath56 .
we examined the implications of these data for imbhs , if any exist in these gcs .
our principal findings are as follows : 1 .
the 206 gcs in m81 lack radio analogs of hlx-1 , a strong imbh candidate in a star cluster in the early - type galaxy eso243 - 49 .
this suggests that hlx-1 is accreting gas related to the 20-myr - old stars in its host cluster or that the hlx-1 phenomenon is so rare that no radio analog is expected in m81 .
two gcs exhibit hard - state x - ray emission . from the empirical fundamental - plane relation ,
their x - ray and radio luminosities suggest individual imbh masses , @xmath36 , of less than @xmath57 and @xmath58 , and associated bh mass fractions , @xmath48 , of less than 0.18 and 0.21 . with only upper limits to the radio luminosities , we can not rule out the likely scenario of contamination from x - ray - emitting stellar - mass bhs in these gcs .
3 . a semi - empirical model developed for milky way gcs converts the upper limits on radio luminosities to upper limits on imbh masses . applying this model to m81 , over a dozen individual gcs appear to have upper limits on the bh mass fractions , @xmath48 , that are below the noteworthy value of 0.15 reported for m60-ucd1 . also , the m81 stacks correspond to imbh masses of @xmath59 for all the gcs and to @xmath52 for the massive gcs .
this model is making inroads into the difficult - to - observe regime of imbhs in extragalactic gcs .
we thank the referee for a helpful and timely report , and dr .
e. greisen for providing the new aips task stack .
jcamj is the recipient of an australian research council future fellowship ( ft140101082 ) .
this research has made use of data obtained from the chandra source catalog , provided by the chandra x - ray center ( cxc ) as part of the chandra data archive .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
any opinions , findings , and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the national science foundation .
mcmullin , j. p. , waters , b. , schiebel , d. , young , w. , & golap , k. 2007 , in asp conf .
376 , astronomical data analysis software and systems , ivi , ed .
r. a. shaw , f. hill , & d. j. bell ( san francisco , ca : asp ) , 127 rrrr 09 55 40.71 & 69 07 19.45 & 2014 jan 5 & [email protected] + 09 54 58.45 & 69 04 43.90 & 2014 jan 6 & [email protected] + 09 56 09.55 & 69 03 20.83 & 2014 jan 5 & [email protected] + 09 55 28.20 & 69 00 51.34 & 2014 jan 6 & [email protected] + rrrr 28 & 0.52e+06 & @xmath600.12e+06 & @xmath600.23 + 31 & 0.43e+06 & @xmath600.11e+06 & @xmath600.26 + 71 & 0.26e+06 & @xmath600.11e+06 & @xmath600.41 + 90 & 0.11e+07 & @xmath600.10e+06 & @xmath600.10 + 100 & 0.28e+06 & @xmath600.10e+06 & @xmath600.37 + 115 & 0.39e+06 & @xmath600.10e+06 & @xmath600.26 + 116 & 0.44e+06 & @xmath600.10e+06 & @xmath600.23 + 118 & 0.19e+07 & @xmath600.10e+06 & @xmath600.05 + 136 & 0.23e+06 & @xmath600.10e+06 & @xmath600.44 + 145 & 0.25e+06 & @xmath600.10e+06 & @xmath600.39 + 158 & 0.53e+06 & @xmath600.10e+06 & @xmath600.19 + 160 & 0.34e+07 & @xmath600.10e+06 & @xmath600.03 + 162 & 0.11e+07 & @xmath600.10e+06 & @xmath600.09 + 175 & 0.45e+06 & @xmath600.10e+06 & @xmath600.23 + 190 & 0.53e+06 & @xmath600.10e+06 & @xmath600.19 + 199 & 0.11e+07 & @xmath600.11e+06 & @xmath600.10 + 201 & 0.23e+06 & @xmath600.10e+06 & @xmath600.43 + 209 & 0.62e+06 & @xmath600.10e+06 & @xmath600.16 + 218 & 0.47e+06 & @xmath600.12e+06 & @xmath600.26 + 226 & 0.22e+06 & @xmath600.10e+06 & @xmath600.47 + 227 & 0.49e+06 & @xmath600.10e+06 & @xmath600.21 + 228 & 0.84e+06 & @xmath600.10e+06 & @xmath600.12 + 232 & 0.65e+06 & @xmath600.11e+06 & @xmath600.17 + 236 & 0.22e+06 & @xmath600.10e+06 & @xmath600.47 + 239 & 0.25e+06 & @xmath600.10e+06 & @xmath600.41 + 246 & 0.15e+07 & @xmath600.10e+06 & @xmath600.07 + 247 & 0.24e+06 & @xmath600.10e+06 & @xmath600.42 + 256 & 0.36e+06 & @xmath600.99e+05 & @xmath600.28 + 269 & 0.16e+07 & @xmath600.10e+06 & @xmath600.07 + 273 & 0.55e+06 & @xmath600.10e+06 & @xmath600.19 + 275 & 0.50e+06 & @xmath600.10e+06 & @xmath600.21 + 277 & 0.43e+06 & @xmath600.99e+05 & @xmath600.23 + 282 & 0.50e+06 & @xmath600.11e+06 & @xmath600.21 + 288 & 0.29e+06 & @xmath600.12e+06 & @xmath600.43 + 292 & 0.54e+06 & @xmath600.99e+05 & @xmath600.18 + 293 & 0.61e+06 & @xmath600.11e+06 & @xmath600.18 + 294 & 0.75e+06 & @xmath600.11e+06 & @xmath600.15 + 295 & 0.23e+06 & @xmath600.11e+06 & @xmath600.46 + 301 & 0.77e+06 & @xmath600.10e+06 & @xmath600.13 + 302 & 0.73e+06 & @xmath600.11e+06 & @xmath600.15 + 304 & 0.23e+06 & @xmath600.10e+06 & @xmath600.45 + 307 & 0.40e+06 & @xmath600.11e+06 & @xmath600.28 + 315 & 0.38e+06 & @xmath600.10e+06 & @xmath600.27 + 330 & 0.24e+06 & @xmath600.11e+06 & @xmath600.45 + 340 & 0.42e+06 & @xmath600.11e+06 & @xmath600.26 + 351 & 0.14e+07 & @xmath600.11e+06 & @xmath600.08 + 388 & 0.56e+06 & @xmath600.11e+06 & @xmath600.20 + 398 & 0.18e+07 & @xmath600.12e+06 & @xmath600.07 + 410 & 0.25e+06 & @xmath600.13e+06 & @xmath600.50 + | nantais et al . used the _ hubble space telescope _ to localize probable globular clusters ( gcs ) in m81 , a spiral galaxy at a distance of 3.63 mpc .
theory predicts that gcs can host intermediate - mass black holes ( imbhs ) with masses @xmath0 .
finding imbhs in gcs could validate a formation channel for seed bhs in the early universe , bolster gravitational - wave predictions for space missions , and test scaling relations between stellar systems and the central bhs they host .
we used the nrao karl g. jansky very large array ( vla ) to search for the radiative signatures of imbh accretion from 206 probable gcs in a mosaic of m81 .
the observing wavelength was 5.5 cm and the spatial resolution was 1.5 ( 26.4 pc ) .
none of the individual gcs are detected , nor are weighted - mean image stacks of the 206 gcs and the 49 massive gcs with stellar masses @xmath1 .
we apply a semi - empirical model to predict the mass of an imbh that , if undergoing accretion in the long - lived hard x - ray state , is consistent with a given radio luminosity .
the 3@xmath2 radio - luminosity upper limits correspond to imbh masses of @xmath3 for the all - cluster stack and @xmath4 for the massive - cluster stack .
we also apply the empirical fundamental - plane relation to two x - ray - detected clusters , finding that their individual imbh masses at 95% confidence are @xmath5 and @xmath6 .
finally , no analog of hlx-1 , a strong imbh candidate in an extragalactic star cluster , occurs in any individual gc in m81 .
this underscores the uniqueness or rarity of the hlx-1 phenomenon . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.